PRACTICAL TOOLKITS & GUIDED EDITIONS

Useful workflows.
Results you can inspect.

Start with CSV reconciliation: inspect how the MIS toolkit traces order and payment mismatches to their source rows using synthetic examples. Compare all 3 runnable toolkits and 108 guided skills below.

Browse 108 guided skills ↓Choose a category pack ↓Watch the MIS example ↗Try free samples ↗

108 guided skills. 3 runnable toolkits. One-time purchases · USD

Read & write reviews ↓

RECORDED SYNTHETIC OUTPUTLOCAL PYTHON RUN

MIS Reporting & Reconciliation

Selected values from examples/expected/summary.json

{
  "counts.orders_reviewed": 9,
  "counts.orders_mismatched": 5,
  "counts.rejected_source_rows": 5
}

The synthetic exports intentionally contain discrepancies. Four of nine orders match; five need investigation. Rejected rows and superseded revisions retain source lineage.

Inspect demo, commands & report ↗

02 / FIND A GUIDED SKILL

Choose the task.
Follow the next steps.

Search the complete collection and preview each task's inputs and outputs. Try a selected free sample before choosing a guided edition.

108 guided skills

Preview first. Choose your guided edition.

01Analysis

Feedback Analysis

Turn customer feedback into evidence-backed themes, mixed sentiment, and a clear action list.

Bring: Feedback records with stable IDs and the business question you want to answer.
Build: Theme and evidence table · Sentiment and limitations · Prioritized actions

Inspect task details

Public task brief · instructional pack · bring your own AI assistant

A task to rehearse

Help me work through feedback analysis. I will supply Feedback records with stable IDs and the business question you want to answer. Produce Theme and evidence table; Sentiment and limitations; Prioritized actions. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.

Check the deliverables

Theme and evidence table: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Sentiment and limitations: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Prioritized actions: trace material statements or calculations to the supplied input; mark anything you cannot verify.

Before you use a result

Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.

This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.

What you get

MIT starter skill, templates, and synthetic worked example
Premium runbook with staged assistant prompts
Premium acceptance checks and usage guide
Single-user premium license and exact file scope

11 files · Guided edition 1.1.0 · License terms · Review this skill

Buy guided edition $0.99Edition notes ↗Try free sample ↗

02Analysis

Data to Story

Create a presentation brief from a dataset while keeping calculations, caveats, and narrative connected.

Bring: A dataset, its metric definitions, the audience, and the decision to support.
Build: Checked metric summary · Slide-by-slide brief · Source and caveat notes

Inspect task details

Public task brief · instructional pack · bring your own AI assistant

A task to rehearse

Help me work through data to story. I will supply A dataset, its metric definitions, the audience, and the decision to support. Produce Checked metric summary; Slide-by-slide brief; Source and caveat notes. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.

Check the deliverables

Checked metric summary: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Slide-by-slide brief: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Source and caveat notes: trace material statements or calculations to the supplied input; mark anything you cannot verify.

Before you use a result

What you get

MIT starter skill, templates, and synthetic worked example
Premium runbook with staged assistant prompts
Premium acceptance checks and usage guide
Single-user premium license and exact file scope

11 files · Guided edition 1.1.0 · License terms · Review this skill

Buy guided edition $0.99Edition notes ↗

03AI design

AI Evaluation Planner

Plan repeatable evaluations for an AI workflow, including supported answers, refusals, and failure cases.

Bring: The AI task, allowed sources, intended users, and the cost of an incorrect answer.
Build: Evaluation matrix · Test cases and rubrics · Release criteria

Inspect task details

Public task brief · instructional pack · bring your own AI assistant

A task to rehearse

Help me work through ai evaluation planner. I will supply The AI task, allowed sources, intended users, and the cost of an incorrect answer. Produce Evaluation matrix; Test cases and rubrics; Release criteria. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.

Check the deliverables

Evaluation matrix: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Test cases and rubrics: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Release criteria: trace material statements or calculations to the supplied input; mark anything you cannot verify.

Before you use a result

What you get

MIT starter skill, templates, and synthetic worked example
Premium runbook with staged assistant prompts
Premium acceptance checks and usage guide
Single-user premium license and exact file scope

11 files · Guided edition 1.1.0 · License terms · Review this skill

Buy guided edition $0.99Edition notes ↗Try free sample ↗

04Automation

SQL Report Review

Review reporting queries for join fan-out, data grain, null handling, and reconciliation gaps.

Bring: A read-only SQL query, database dialect, table schemas, and expected reporting grain.
Build: Prioritized findings · Suggested read-only query · Reconciliation checks

Inspect task details

Public task brief · instructional pack · bring your own AI assistant

A task to rehearse

Help me work through sql report review. I will supply A read-only SQL query, database dialect, table schemas, and expected reporting grain. Produce Prioritized findings; Suggested read-only query; Reconciliation checks. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.

Check the deliverables

Prioritized findings: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Suggested read-only query: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Reconciliation checks: trace material statements or calculations to the supplied input; mark anything you cannot verify.

Before you use a result

What you get

MIT starter skill, templates, and synthetic worked example
Premium runbook with staged assistant prompts
Premium acceptance checks and usage guide
Single-user premium license and exact file scope

12 files · Guided edition 1.1.0 · License terms · Review this skill

Buy guided edition $0.99Edition notes ↗Inspect worked preview ↗Try free sample ↗

05Automation

VBA Automation Planner

Turn a recurring spreadsheet process into a scoped automation plan with dry runs, reconciliation, and recovery.

Bring: Workbook structure, manual steps, expected outputs, and known exceptions.
Build: Automation brief · Validation and dry-run plan · Recovery and handover checklist

Inspect task details

Public task brief · instructional pack · bring your own AI assistant

A task to rehearse

Help me work through vba automation planner. I will supply Workbook structure, manual steps, expected outputs, and known exceptions. Produce Automation brief; Validation and dry-run plan; Recovery and handover checklist. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.

Check the deliverables

Automation brief: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Validation and dry-run plan: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Recovery and handover checklist: trace material statements or calculations to the supplied input; mark anything you cannot verify.

Before you use a result

What you get

MIT starter skill, templates, and synthetic worked example
Premium runbook with staged assistant prompts
Premium acceptance checks and usage guide
Single-user premium license and exact file scope

11 files · Guided edition 1.1.0 · License terms · Review this skill

Buy guided edition $0.99Edition notes ↗

06AI design

Agent Workflow Designer

Design a bounded agent workflow with clear state transitions, tool contracts, and human handoffs.

Bring: The goal, available tools, data sources, and which actions need human decisions.
Build: Workflow specification · Tool and state contracts · Failure and handoff scenarios

Inspect task details

Public task brief · instructional pack · bring your own AI assistant

A task to rehearse

Help me work through agent workflow designer. I will supply The goal, available tools, data sources, and which actions need human decisions. Produce Workflow specification; Tool and state contracts; Failure and handoff scenarios. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.

Check the deliverables

Workflow specification: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Tool and state contracts: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Failure and handoff scenarios: trace material statements or calculations to the supplied input; mark anything you cannot verify.

Before you use a result

What you get

MIT starter skill, templates, and synthetic worked example
Premium runbook with staged assistant prompts
Premium acceptance checks and usage guide
Single-user premium license and exact file scope

11 files · Guided edition 1.1.0 · License terms · Review this skill

Buy guided edition $0.99Edition notes ↗

More skills in the collection

Every skill is listed here when JavaScript is off. Use the links to inspect or purchase a guided edition.

Metric Definition · Analysis
Turn an ambiguous KPI into a precise, reproducible metric contract.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through metric definition. I will supply The business decision, event grain, reporting window, and inclusion rules. Produce Metric contract; Worked reconciliation; Open definition decisions. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Metric contract: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Worked reconciliation: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Open definition decisions: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Cohort Retention · Analysis
Compare cohort retention without treating unobserved future periods as churn.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through cohort retention. I will supply Cohort membership, return events, observation cutoff, and retention definition. Produce Cohort retention table; Eligibility and censoring notes; Interpretation brief. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Cohort retention table: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Eligibility and censoring notes: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Interpretation brief: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Funnel Diagnostics · Analysis
Find genuine funnel drop-offs while excluding out-of-order events.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through funnel diagnostics. I will supply Timestamped events, entity keys, ordered funnel stages, and a conversion window. Produce Ordered funnel table; Drop-off evidence; Tracking and investigation notes. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Ordered funnel table: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Drop-off evidence: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Tracking and investigation notes: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Experiment Readout · Analysis
Produce an experiment readout with effect sizes and release guardrails.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through experiment readout. I will supply Assignment counts, metric outcomes, analysis plan, and guardrail thresholds. Produce Effect-size readout; Design and validity checks; Decision recommendation. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Effect-size readout: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Design and validity checks: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Decision recommendation: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Forecast Backtest · Analysis
Benchmark a forecast against a baseline with auditable error calculations.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through forecast backtest. I will supply Forecast origins, horizons, actuals, model predictions, and a baseline. Produce Aligned error table; Metric and bias comparison; Backtest limitations. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Aligned error table: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Metric and bias comparison: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Backtest limitations: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Anomaly Triage · Analysis
Separate a real metric incident from partial data or reporting artifacts.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through anomaly triage. I will supply Observed values, expected baseline, completeness evidence, and relevant changes. Produce Triage assessment; Evidence and competing hypotheses; Investigation order. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Triage assessment: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Evidence and competing hypotheses: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Investigation order: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Dashboard Brief · Analysis
Design a dashboard around decisions, denominators, and actionable exceptions.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through dashboard brief. I will supply Audience decisions, available data, metric contracts, and refresh needs. Produce Dashboard specification; Metric and interaction contracts; Acceptance scenarios. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Dashboard specification: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Metric and interaction contracts: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Acceptance scenarios: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
KPI Reconciliation · Analysis
Explain KPI disagreements with a signed, evidence-backed reconciliation bridge.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through kpi reconciliation. I will supply Two reported values, their definitions, and attributable adjustment records. Produce Signed reconciliation bridge; Residual and evidence table; Definition alignment notes. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Signed reconciliation bridge: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Residual and evidence table: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Definition alignment notes: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗Inspect worked preview ↗ Review this skill
Segment Comparison · Analysis
Compare like-for-like segment results and expose misleading aggregate rankings.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through segment comparison. I will supply Segment counts and outcomes, group definitions, and comparison objectives. Produce Within-segment comparison; Aggregate and adjusted results; Mix and inference caveats. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Within-segment comparison: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Aggregate and adjusted results: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Mix and inference caveats: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Missing Data Assessment · Analysis
Identify how missing observations affect coverage and conclusions.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through missing data assessment. I will supply Field-level missingness, observed values, reasons, and valid value bounds. Produce Missingness profile; Observed-only results; Sensitivity and follow-up plan. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Missingness profile: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Observed-only results: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Sensitivity and follow-up plan: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Trend and Seasonality · Analysis
Read time trends without confusing recurring seasonality with growth.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through trend and seasonality. I will supply A dated series, calendar conventions, complete-period flags, and comparison needs. Produce Comparable-period calculations; Trend and seasonal interpretation; Evidence limitations. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Comparable-period calculations: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Trend and seasonal interpretation: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Evidence limitations: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Pricing Sensitivity · Analysis
Compare price scenarios while separating margin arithmetic from demand assumptions.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through pricing sensitivity. I will supply Price, expected units, variable costs, and scenario assumptions. Produce Scenario contribution table; Break-even volume threshold; Assumption and inference notes. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Scenario contribution table: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Break-even volume threshold: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Assumption and inference notes: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Inventory Coverage · Analysis
Assess stock coverage using available inventory and explicit replenishment timing.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through inventory coverage. I will supply SKU inventory, reservations, demand rates, and replenishment lead times. Produce Coverage and exposure table; Lead-time gap assessment; Inventory assumptions. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Coverage and exposure table: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Lead-time gap assessment: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Inventory assumptions: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Support Capacity · Analysis
Model support workload and usable capacity with explicit shrinkage assumptions.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through support capacity. I will supply Ticket arrivals, handling minutes, staffing hours, shrinkage, and backlog targets. Produce Workload and capacity model; Gap and backlog scenarios; Queueing limitations. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Workload and capacity model: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Gap and backlog scenarios: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Queueing limitations: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Scenario Model · Analysis
Build internally consistent scenarios instead of mixing incompatible assumptions.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through scenario model. I will supply Driver definitions, linked assumptions, costs, and the decision threshold. Produce Driver-based scenario table; Break-even calculation; Sensitivity and assumptions. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Driver-based scenario table: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Break-even calculation: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Sensitivity and assumptions: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Chart Selection · Analysis
Choose visuals that make the analytical comparison accurate and clear.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through chart selection. I will supply The audience question, dataset shape, units, and uncertainty information. Produce Chart recommendation; Encoding and annotation specification; Misleading-chart checks. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Chart recommendation: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Encoding and annotation specification: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Misleading-chart checks: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Executive Insight Brief · Analysis
Turn reconciled findings into a concise executive decision brief.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through executive insight brief. I will supply Verified findings, metric definitions, decision options, and material caveats. Produce Executive brief; Evidence-to-decision mapping; Decision request and caveats. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Executive brief: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Evidence-to-decision mapping: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Decision request and caveats: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Survey Question Review · Customer intelligence
Improve survey questions so answers measure one clearly defined construct.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through survey question review. I will supply Survey questions, target respondents, measurement intent, and answer scales. Produce Item-level findings; Rewritten survey items; Measurement trade-offs. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Item-level findings: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Rewritten survey items: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Measurement trade-offs: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
NPS Driver Review · Customer intelligence
Calculate NPS correctly and distinguish observed themes from causal drivers.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through nps driver review. I will supply Valid 0-10 recommendation scores, linked comments, and sampling context. Produce NPS reconciliation; Score-group theme table; Driver hypotheses and limitations. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- NPS reconciliation: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Score-group theme table: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Driver hypotheses and limitations: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Journey Friction Map · Customer intelligence
Connect journey-stage friction with affected customers and traceable evidence.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through journey friction map. I will supply Journey stages, customer-linked touchpoints, outcomes, and observation limits. Produce Stage-level friction map; Customer and touchpoint counts; Investigation priorities. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Stage-level friction map: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Customer and touchpoint counts: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Investigation priorities: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Churn Signal Review · Customer intelligence
Review churn-warning rules without leaking future evidence into decisions.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through churn signal review. I will supply Customer behavior before a cutoff, an explicit rule, and outcome definitions. Produce Cutoff-valid signal table; Leakage and coverage findings; Validation requirements. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Cutoff-valid signal table: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Leakage and coverage findings: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Validation requirements: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Contact Reason Taxonomy · Customer intelligence
Create contact labels that remain consistent when tickets contain several issues.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through contact reason taxonomy. I will supply Representative ticket text, reporting purpose, and existing label rules. Produce Taxonomy and definitions; Labeled case table; Boundary and review rules. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Taxonomy and definitions: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Labeled case table: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Boundary and review rules: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Ticket Priority Rubric · Customer intelligence
Assign transparent ticket priority using operational impact and urgency.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through ticket priority rubric. I will supply Ticket impact, urgency, available workaround, and a stated priority policy. Produce Priority rubric; Ticket decision table; Ambiguity and escalation notes. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Priority rubric: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Ticket decision table: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Ambiguity and escalation notes: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Service Recovery Plan · Customer intelligence
Plan an empathetic recovery with a bounded remedy and clear follow-up.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through service recovery plan. I will supply Verified issue, customer impact, remedy policy, and available owner roles. Produce Recovery action plan; Customer response draft; Policy exceptions and follow-up. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Recovery action plan: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Customer response draft: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Policy exceptions and follow-up: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Voice of Customer Brief · Customer intelligence
Create a customer-voice brief with honest scope and traceable recommendations.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through voice of customer brief. I will supply Customer comments, collection context, duplicate rules, and the target decision. Produce Customer voice brief; Evidence and sample accounting; Prioritized investigation proposals. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Customer voice brief: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Evidence and sample accounting: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Prioritized investigation proposals: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Review Comparison · Customer intelligence
Compare review evidence without treating unequal samples as population rankings.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through review comparison. I will supply Review records, rating scales, source populations, and comparison questions. Produce Source-level comparison; Aspect evidence table; Comparability and sampling caveats. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Source-level comparison: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Aspect evidence table: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Comparability and sampling caveats: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Feedback Deduplication · Customer intelligence
Clean feedback without merging independent voices or losing revisions.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through feedback deduplication. I will supply Feedback IDs, source keys, revision timestamps or versions, and exclusion rules. Produce Accepted record table; Duplicate and revision ledger; Count reconciliation. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Accepted record table: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Duplicate and revision ledger: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Count reconciliation: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Aspect Sentiment · Customer intelligence
Keep positive and negative sentiment attached to the correct aspect.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through aspect sentiment. I will supply Text records, aspect definitions, annotation rules, and reporting unit. Produce Aspect-level annotation table; Sentiment count summary; Ambiguity and evidence notes. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Aspect-level annotation table: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Sentiment count summary: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Ambiguity and evidence notes: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗Inspect worked preview ↗ Review this skill
Complaint Root Cause · Customer intelligence
Separate repeated complaint symptoms from verified underlying causes.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through complaint root cause. I will supply Complaints, linked logs or process evidence, and candidate explanations. Produce Symptom and cause evidence map; Confirmed versus hypothesized findings; Next diagnostic checks. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Symptom and cause evidence map: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Confirmed versus hypothesized findings: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Next diagnostic checks: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Interview Synthesis · Customer intelligence
Synthesize interview evidence without inflating prevalence from repeated quotes.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through interview synthesis. I will supply Interview excerpts, participant IDs, study question, and sampling context. Produce Theme and participant matrix; Evidence-linked synthesis; Divergence and follow-up questions. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Theme and participant matrix: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Evidence-linked synthesis: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Divergence and follow-up questions: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Persona Evidence · Customer intelligence
Create behavior-based personas with explicit evidence and uncertainty.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through persona evidence. I will supply Participant observations, goals, constraints, and the intended design decision. Produce Provisional persona profiles; Attribute evidence table; Validation and gaps. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Provisional persona profiles: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Attribute evidence table: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Validation and gaps: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Product Feedback Roadmap · Customer intelligence
Prioritize product opportunities while keeping constraints and scoring assumptions visible.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through product feedback roadmap. I will supply Feedback evidence, estimated reach and impact, effort, confidence, and mandatory constraints. Produce Candidate roadmap ranking; Scoring and constraint ledger; Discovery and measurement plan. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Candidate roadmap ranking: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Scoring and constraint ledger: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Discovery and measurement plan: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
CX Metric Audit · Customer intelligence
Check CX metrics for denominator errors and misleading comparisons.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through cx metric audit. I will supply Survey questions, response counts, metric definitions, and reporting claims. Produce Metric audit findings; Corrected calculations; Comparability and coverage notes. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Metric audit findings: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Corrected calculations: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Comparability and coverage notes: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Escalation Pattern Review · Customer intelligence
Find escalation patterns without overcounting transfers or treating open tickets as resolved.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through escalation pattern review. I will supply Ticket histories, escalation events, resolved times, and a reporting cutoff. Produce Escalation pattern table; Ticket and event reconciliation; Timing and case-mix caveats. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Escalation pattern table: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Ticket and event reconciliation: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Timing and case-mix caveats: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Prompt Contract · AI design
Turn a vague prompt into explicit input rules, evidence boundaries, and observable acceptance checks.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through prompt contract. I will supply The user task, sample inputs, authorized evidence, output consumers, and failure consequences. Produce Prompt contract; Ambiguity register; Boundary acceptance cases. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Prompt contract: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Ambiguity register: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Boundary acceptance cases: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Structured Output Design · AI design
Design response schemas that distinguish unknown values from negative results and invalid output.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through structured output design. I will supply Consumer requirements, sample responses, field meanings, and known invalid combinations. Produce Response schema brief; Cross-field invariants; Valid and invalid examples. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Response schema brief: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Cross-field invariants: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Valid and invalid examples: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Refusal Policy Design · AI design
Define when an assistant answers, clarifies, abstains, or hands a request to a person.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through refusal policy design. I will supply Assistant remit, permitted actions, supplied restrictions, and representative boundary requests. Produce Decision policy; Reason codes; Boundary response examples. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Decision policy: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Reason codes: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Boundary response examples: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Tool Contract Review · AI design
Review tool inputs and outcomes so retries and ambiguous failures cannot silently duplicate actions.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through tool contract review. I will supply Tool specification, examples, authorization model, error cases, and side-effect behavior. Produce Contract findings; Revised specification; Retry and ambiguity cases. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Contract findings: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Revised specification: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Retry and ambiguity cases: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Agent State Machine · AI design
Turn an agent workflow into explicit states, guarded transitions, and stale-event handling.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through agent state machine. I will supply Workflow goal, observable events, action permissions, failure paths, and state persistence assumptions. Produce State transition table; Invariant register; Event trace review. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- State transition table: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Invariant register: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Event trace review: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Agent Memory Plan · AI design
Design memory records with provenance, expiry, correction, and clear user-control boundaries.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through agent memory plan. I will supply Memory use cases, proposed fields, source trust, retention requirements, and correction examples. Produce Memory schema; Retention and correction rules; Retrieval boundary cases. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Memory schema: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Retention and correction rules: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Retrieval boundary cases: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Human Handoff Design · AI design
Design handoffs with the decision needed, evidence, ownership, and safe resume conditions.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through human handoff design. I will supply Escalation triggers, available human roles, workflow state, evidence, and communication permissions. Produce Handoff packet; Ownership and timeout rules; Resume decision cases. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Handoff packet: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Ownership and timeout rules: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Resume decision cases: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Hallucination Test Suite · AI design
Test unsupported details and false certainty with evidence-based expected answers.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through hallucination test suite. I will supply Assistant scope, approved source snapshot, candidate answers, and claim severity. Produce Grounding case suite; Claim-level answer keys; Failure severity report. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Grounding case suite: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Claim-level answer keys: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Failure severity report: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Prompt Injection Test Plan · AI design
Test whether untrusted content can redirect answers, reveal protected instructions, or trigger actions.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through prompt injection test plan. I will supply Task instructions, untrusted entry points, tool capabilities, and expected trust boundaries. Produce Threat-path matrix; Synthetic injection cases; Expected boundary outcomes. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Threat-path matrix: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Synthetic injection cases: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Expected boundary outcomes: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
LLM Cost Budget · AI design
Calculate token budgets and cost drivers with explicit retry and fallback assumptions.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through llm cost budget. I will supply Expected request volume, token profiles, supplied unit rates, retries, fallbacks, and budget limits. Produce Cost calculation table; Sensitivity scenarios; Budget guardrail proposal. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Cost calculation table: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Sensitivity scenarios: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Budget guardrail proposal: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Model Routing Plan · AI design
Route tasks using segment-level evidence and explicit escalation criteria.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through model routing plan. I will supply Task strata, supplied model results, resource costs, latency limits, and quality floors. Produce Routing decision table; Fallback criteria; Evidence and uncertainty notes. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Routing decision table: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Fallback criteria: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Evidence and uncertainty notes: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
AI Incident Triage · AI design
Build an evidence-based incident timeline and prioritize containment without premature root-cause claims.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through ai incident triage. I will supply Failure report, supplied traces, affected tasks, recent changes, and operating authority. Produce Impact and timeline brief; Cause hypotheses; Containment and validation plan. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Impact and timeline brief: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Cause hypotheses: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Containment and validation plan: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
AI Use Case Scope · AI design
Turn a broad AI idea into a bounded pilot with measurable benefit and explicit exclusions.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through ai use case scope. I will supply User problem, workflow baseline, candidate tasks, evidence availability, and decision constraints. Produce Pilot scope brief; Benefit hypothesis; Acceptance and exclusion criteria. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Pilot scope brief: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Benefit hypothesis: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Acceptance and exclusion criteria: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
AI Release Review · AI design
Make a release recommendation that respects critical gates and untested conditions.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through ai release review. I will supply Release candidate changes, frozen evaluation results, acceptance gates, and operating constraints. Produce Gate decision; Failed-case priorities; Required evidence before release. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Gate decision: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Failed-case priorities: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Required evidence before release: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Synthetic Test Data · AI design
Create reproducible synthetic cases with constraints, edge labels, and explicit limitations.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through synthetic test data. I will supply Task schema, domain constraints, coverage targets, and permitted synthetic variations. Produce Synthetic fixture specification; Expected outcome labels; Coverage and constraint checks. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Synthetic fixture specification: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Expected outcome labels: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Coverage and constraint checks: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
AI Observability Plan · AI design
Design traces and metrics that separate pipeline failures while keeping data collection purposeful.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through ai observability plan. I will supply Workflow stages, failure questions, existing telemetry, data sensitivity, and operational thresholds. Produce Trace and metric schema; Failure attribution map; Retention and alert proposal. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Trace and metric schema: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Failure attribution map: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Retention and alert proposal: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Response Quality Rubric · AI design
Score answers against observable criteria and critical failures rather than writing style alone.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through response quality rubric. I will supply Task objective, source evidence, candidate responses, consequence levels, and reviewer disagreements. Produce Anchored scoring rubric; Scored examples; Disagreement resolution rules. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Anchored scoring rubric: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Scored examples: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Disagreement resolution rules: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
RAG Source Readiness · NLP & retrieval
Check whether a source collection can support trustworthy retrieval before indexing it.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through rag source readiness. I will supply Source inventory, sample extracted text, authority rules, revision history, and intended questions. Produce Source readiness matrix; Blocking evidence gaps; Indexing preparation plan. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Source readiness matrix: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Blocking evidence gaps: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Indexing preparation plan: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Chunking Experiment · NLP & retrieval
Compare chunk designs using answer evidence and boundary coverage rather than chunk size alone.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through chunking experiment. I will supply Document excerpts, question set, answer spans, candidate chunk boundaries, and retrieval budget. Produce Chunk comparison plan; Evidence coverage findings; Boundary and duplication cases. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Chunk comparison plan: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Evidence coverage findings: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Boundary and duplication cases: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Retrieval Evaluation · NLP & retrieval
Calculate retrieval metrics while keeping answerability and incomplete judgments visible.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through retrieval evaluation. I will supply Queries, ranked source IDs, judged relevant sets, cutoff k, and label completeness. Produce Per-query retrieval metrics; Aggregate and denominator notes; Error analysis. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Per-query retrieval metrics: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Aggregate and denominator notes: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Error analysis: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗Inspect worked preview ↗ Review this skill
Citation Check · NLP & retrieval
Audit claim-to-source support beyond whether a citation link exists.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through citation check. I will supply Answer text, cited excerpts, source dates, and claim-level support requirements. Produce Claim-citation matrix; Unsupported claim findings; Corrected answer draft. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Claim-citation matrix: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Unsupported claim findings: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Corrected answer draft: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Query Rewrite Review · NLP & retrieval
Improve search wording while preserving the user's actual constraints and uncertainty.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through query rewrite review. I will supply Original query, candidate rewrites, corpus terminology, and required filters. Produce Rewrite fidelity findings; Suggested query variants; Constraint preservation checks. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Rewrite fidelity findings: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Suggested query variants: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Constraint preservation checks: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Hybrid Search Plan · NLP & retrieval
Plan a hybrid search experiment that protects exact identifiers and semantic intent.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through hybrid search plan. I will supply Query classes, sample lexical/dense rankings, corpus metadata, filters, and evaluation labels. Produce Retrieval branch plan; Fusion and filter rules; Comparison cases. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Retrieval branch plan: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Fusion and filter rules: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Comparison cases: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Knowledge Base Refresh · NLP & retrieval
Plan source refreshes that retire stale chunks and verify the active evidence snapshot.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through knowledge base refresh. I will supply Source change manifest, indexed versions, deletion records, freshness requirements, and release controls. Produce Refresh manifest; Reconciliation checks; Activation and recovery plan. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Refresh manifest: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Reconciliation checks: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Activation and recovery plan: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Entity Extraction Schema · NLP & retrieval
Extract entities without merging names, roles, or dates beyond what the text supports.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through entity extraction schema. I will supply Text corpus sample, target entity types, normalization needs, and ambiguity rules. Produce Entity schema; Span-backed extraction examples; Ambiguity and validation rules. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Entity schema: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Span-backed extraction examples: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Ambiguity and validation rules: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Document Classification · NLP & retrieval
Classify documents using explicit label boundaries and mixed-content rules.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through document classification. I will supply Label definitions, document examples, desired label cardinality, and review costs. Produce Label decision rules; Evidence-backed classifications; Ambiguity review queue. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Label decision rules: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Evidence-backed classifications: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Ambiguity review queue: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Text Deduplication · NLP & retrieval
Deduplicate repeated records without erasing separate events or contradictory versions.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through text deduplication. I will supply Records with stable IDs, event keys, text, provenance, and version or timing information. Produce Duplicate group table; Conflict and retention rules; Adjusted counting notes. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Duplicate group table: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Conflict and retention rules: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Adjusted counting notes: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Multilingual Evaluation · NLP & retrieval
Review multilingual quality without letting a larger language sample hide a failing segment.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through multilingual evaluation. I will supply Language-tagged cases, original meanings, candidate outputs, scoring rules, and language-specific gates. Produce Language-level scores; Meaning-error analysis; Coverage gaps. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Language-level scores: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Meaning-error analysis: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Coverage gaps: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Taxonomy Design · NLP & retrieval
Design theme hierarchies with clear boundaries and evidence-aware rollup rules.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through taxonomy design. I will supply Business questions, sample feedback, proposed themes, hierarchy constraints, and counting grain. Produce Taxonomy and label guide; Assignment examples; Rollup and revision rules. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Taxonomy and label guide: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Assignment examples: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Rollup and revision rules: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Summarization Fidelity · NLP & retrieval
Summarize evidence while preserving counts, uncertainty, and unresolved outcomes.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through summarization fidelity. I will supply Source excerpts, audience, compression target, and required factual details. Produce Faithful summary; Claim-to-source check; Omitted-detail and caveat notes. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Faithful summary: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Claim-to-source check: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Omitted-detail and caveat notes: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
PII Redaction Review · NLP & retrieval
Review redaction for direct identifiers and useful context without promising anonymization.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through pii redaction review. I will supply Synthetic or approved text, sharing purpose, sensitive-field policy, and proposed redaction. Produce Redaction findings; Minimized text draft; Linkability and residual-risk notes. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Redaction findings: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Minimized text draft: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Linkability and residual-risk notes: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Semantic Search Debug · NLP & retrieval
Locate retrieval failures in the pipeline before proposing embedding or ranking changes.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through semantic search debug. I will supply Original queries, rewritten queries, filters, candidate lists, source state, and expected evidence. Produce Failure-stage diagnosis; Discriminating checks; Prioritized repair proposal. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Failure-stage diagnosis: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Discriminating checks: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Prioritized repair proposal: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Retrieval Ablation · NLP & retrieval
Compare retrieval components with paired cases and explicit configuration controls.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through retrieval ablation. I will supply Baseline and ablation configurations, fixed queries, source snapshot, and per-query outcomes. Produce Ablation comparison; Paired regression findings; Follow-up experiment plan. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Ablation comparison: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Paired regression findings: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Follow-up experiment plan: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Evidence Conflict Review · NLP & retrieval
Resolve conflicting evidence by authority and scope while preserving unresolved cases.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through evidence conflict review. I will supply Contradictory claims, source excerpts, authority hierarchy, dates, and the question's time scope. Produce Conflict matrix; Supported resolution or abstention; Evidence follow-up questions. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Conflict matrix: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Supported resolution or abstention: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Evidence follow-up questions: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Data Contract · Data engineering
Define the rules a dataset must satisfy before downstream consumers can rely on it.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through data contract. I will supply Dataset purpose, field definitions, consumer needs, and sample records. Produce Dataset contract; Acceptance rules; Change decisions. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Dataset contract: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Acceptance rules: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Change decisions: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Schema Drift Review · Data engineering
Identify structural and semantic changes before consuming a new schema.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through schema drift review. I will supply Old/new schemas, sample values, and consumer assumptions. Produce Change classification; Consumer impact map; Compatibility checks. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Change classification: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Consumer impact map: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Compatibility checks: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Incremental Load Plan · Data engineering
Plan incremental loads that account for timestamp ties, corrections, and late arrivals.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through incremental load plan. I will supply Change model, keys, timestamps, and target merge behavior. Produce Load boundaries; Checkpoint rules; Replay cases. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Load boundaries: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Checkpoint rules: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Replay cases: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Pipeline Reconciliation · Data engineering
Explain mismatches with key-level evidence and conserved counts or amounts.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through pipeline reconciliation. I will supply Source/target extracts, keys, filters, and permitted transformations. Produce Reconciliation table; Mismatch classification; Follow-up checks. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Reconciliation table: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Mismatch classification: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Follow-up checks: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Data Lineage Map · Data engineering
Trace report fields back to source evidence and identify incomplete lineage.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through data lineage map. I will supply Source schemas, transformations, output fields, and dependency boundaries. Produce Dependency map; Field lineage table; Unresolved provenance questions. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Dependency map: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Field lineage table: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Unresolved provenance questions: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Data Quality Rules · Data engineering
Build measurable quality checks without double-counting multi-rule failures.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through data quality rules. I will supply Grain, constraints, sample records, and release policy. Produce Rule catalog; Exception table; Batch acceptance summary. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Rule catalog: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Exception table: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Batch acceptance summary: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Backfill Plan · Data engineering
Design historical reruns that preserve current data and prove replacement windows.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through backfill plan. I will supply Historical scope, source availability, target grain, and recovery constraints. Produce Backfill windows; Replacement/checkpoint rules; Reconciliation/recovery checks. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Backfill windows: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Replacement/checkpoint rules: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Reconciliation/recovery checks: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Event Deduplication · Data engineering
Separate repeated delivery from legitimate events and conflicting payloads.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through event deduplication. I will supply Event IDs, payloads, immutability/version semantics, and retention rules. Produce Duplicate policy; Classified event ledger; Replay/conflict checks. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Duplicate policy: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Classified event ledger: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Replay/conflict checks: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Slowly Changing Dimensions · Data engineering
Plan dimension history so facts use the right attributes at the right time.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through slowly changing dimensions. I will supply Dimension keys, attributes, effective dates, and correction rules. Produce History strategy; Effective intervals; As-of join checks. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- History strategy: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Effective intervals: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- As-of join checks: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
SQL Window Review · Data engineering
Check analytical SQL for tied ordering, wrong frames, and unstable selection.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through sql window review. I will supply Dialect, query, partition grain, metric, and tie examples. Produce Window findings; Read-only expressions; Expected row outputs. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Window findings: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Read-only expressions: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Expected row outputs: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
SQL Performance Triage · Data engineering
Prioritize bottlenecks from evidence without promising untested speedups.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through sql performance triage. I will supply Query, dialect, actual/estimated plan, timings, and correctness constraints. Produce Ranked bottlenecks; Controlled comparison plan; Correctness guardrails. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Ranked bottlenecks: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Controlled comparison plan: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Correctness guardrails: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Data Access Matrix · Data engineering
Map the minimum data each role needs for a defined workflow.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through data access matrix. I will supply Roles, purposes, fields, row scopes, and supplied sensitivity rules. Produce Role/field matrix; Scope/retention notes; Owner decisions. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Role/field matrix: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Scope/retention notes: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Owner decisions: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Data Migration Checklist · Data engineering
Plan migration proof from source mapping through cutover readiness.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through data migration checklist. I will supply Source/target schemas, scope, transformations, and cutover constraints. Produce Migration checklist; Expected reconciliations; Cutover/recovery criteria. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Migration checklist: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Expected reconciliations: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Cutover/recovery criteria: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
API Data Mapping · Data engineering
Translate nested API samples without accidental row multiplication.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through api data mapping. I will supply Responses, target grain, semantics, and known pagination/status rules. Produce Field mapping; Array/grain decisions; Sample target rows. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Field mapping: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Array/grain decisions: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Sample target rows: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Batch Failure Triage · Data engineering
Identify whether a failed batch can retry safely from committed-state evidence.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through batch failure triage. I will supply Manifest, logs, stage boundaries, commit evidence, and retry semantics. Produce Failure timeline; Recovery options; Missing evidence/stop conditions. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Failure timeline: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Recovery options: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Missing evidence/stop conditions: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Timestamp Normalization · Data engineering
Align timestamps consistently while exposing missing timezone or format assumptions.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through timestamp normalization. I will supply Timestamp values, locale, offset rules, and reporting boundary. Produce Normalized table; Ambiguity exceptions; Boundary checks. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Normalized table: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Ambiguity exceptions: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Boundary checks: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Analytics Table Grain · Data engineering
Make grain explicit so reports count each intended fact once.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through analytics table grain. I will supply Table samples, analytical questions, keys, and measure definitions. Produce Grain/key statement; Aggregation rules; Join-risk examples. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Grain/key statement: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Aggregation rules: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Join-risk examples: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Spreadsheet Reconciliation · Automation
Reconcile spreadsheet exports without hiding missing rows behind matching totals.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through spreadsheet reconciliation. I will supply Two sheet extracts, stable keys, formula/value rules, and rounding policy. Produce Matched/exception table; Adjustment bridge; Workbook review checks. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Matched/exception table: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Adjustment bridge: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Workbook review checks: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Workbook Input Validation · Automation
Validate workbook inputs before a reporting or automation process uses them.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through workbook input validation. I will supply Sheet/table layout, field rules, locale, and representative rows. Produce Input contract; Row exceptions; Acceptance criteria. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Input contract: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Row exceptions: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Acceptance criteria: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Report Scheduling Plan · Automation
Schedule reports by data readiness and business deadlines rather than clock time alone.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through report scheduling plan. I will supply Refresh deadlines, timezone, upstream readiness, recipients, and failure rules. Produce Scheduling design; Dependency gates; Failure/delivery scenarios. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Scheduling design: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Dependency gates: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Failure/delivery scenarios: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Automation Opportunity Review · Automation
Choose automation candidates using evidence of effort and process stability.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through automation opportunity review. I will supply Candidate processes, frequencies, effort, exceptions, and constraints. Produce Candidate comparison; Priority rationale; Pilot acceptance criteria. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Candidate comparison: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Priority rationale: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Pilot acceptance criteria: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Process Mapping · Automation
Make process handoffs and exception paths explicit before changing them.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through process mapping. I will supply Current steps, actors, inputs, handoffs, timings, and exceptions. Produce Current-state map; Exception paths; Bottleneck questions. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Current-state map: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Exception paths: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Bottleneck questions: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Exception Routing · Automation
Route failed or ambiguous records to the right decision owner.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through exception routing. I will supply Exception types, evidence, impact, owners, and resumption rules. Produce Routing table; Exception payload; Resolution/resumption cases. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Routing table: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Exception payload: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Resolution/resumption cases: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Idempotency Review · Automation
Identify repeated-run effects and define stable operation identities.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through idempotency review. I will supply Workflow steps, target effects, key/version rules, and failure boundaries. Produce Repeatability findings; Operation identity design; Crash/replay scenarios. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Repeatability findings: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Operation identity design: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Crash/replay scenarios: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Automation Handover · Automation
Create an operating handover that another owner can use without hidden assumptions.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through automation handover. I will supply Workflow design, operational evidence, dependencies, owners, and known limits. Produce Handover brief; Run interpretation guide; Open ownership questions. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Handover brief: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Run interpretation guide: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Open ownership questions: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
File Ingestion Plan · Automation
Plan reliable file intake with visible completeness and replay rules.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through file ingestion plan. I will supply File naming, manifest/schema rules, versions, targets, and failure behavior. Produce Ingestion contract; File classification; Commit/replay checks. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Ingestion contract: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- File classification: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Commit/replay checks: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Email Draft Workflow · Automation
Prepare repeatable email drafts with verified context and review boundaries.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through email draft workflow. I will supply Purpose, recipients, approved source facts, tone, and review rules. Produce Draft workflow; Message draft; Recipient/fact checks. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Draft workflow: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Message draft: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Recipient/fact checks: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Approval Workflow · Automation
Design clear approval transitions and prevent stale decisions from authorizing revised work.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through approval workflow. I will supply Approval purpose, authority rules, versions, and deadlines. Produce State-transition table; Approval evidence contract; Revision/expiry scenarios. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- State-transition table: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Approval evidence contract: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Revision/expiry scenarios: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Document Template Plan · Automation
Specify reusable documents without fragile find-and-replace assumptions.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through document template plan. I will supply Document purpose, fields, sample records, conditions, and presentation rules. Produce Template specification; Binding/conditional rules; Sample document outline. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Template specification: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Binding/conditional rules: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Sample document outline: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Automation Regression Plan · Automation
Plan regression checks that protect observable behavior across automation changes.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through automation regression plan. I will supply Previous behavior, proposed change, input cases, and failure boundaries. Produce Risk-based test matrix; Expected observable results; Release evidence gaps. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Risk-based test matrix: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Expected observable results: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Release evidence gaps: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Recurring Report Pack · Automation
Create repeatable reporting packs whose numbers, narrative, and caveats agree.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through recurring report pack. I will supply Audience, reporting period, metrics, sources, and recurring sections. Produce Report-pack outline; Metric/commentary consistency; Correction/version rules. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Report-pack outline: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Metric/commentary consistency: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Correction/version rules: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Workflow Retry Review · Automation
Bound retries to recoverable failures and avoid repeating uncertain external effects.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through workflow retry review. I will supply Stages, error classes, attempt limits, effects, and commit evidence. Produce Retry decision table; Attempt timeline; Stop/reconciliation rules. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Retry decision table: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Attempt timeline: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Stop/reconciliation rules: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Automation Value Estimate · Automation
Calculate net time potential and break-even assumptions transparently.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through automation value estimate. I will supply Frequency, time measurements, review/maintenance estimates, setup effort, and value basis. Produce Effort model; Break-even calculation; Sensitivity and evidence gaps. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Effort model: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Break-even calculation: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Sensitivity and evidence gaps: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill
Operating Procedure Draft · Automation
Write a usable procedure with clear inputs, stop conditions, and completion evidence.
Inspect task details
Public task brief · instructional pack · bring your own AI assistant
A task to rehearse
Help me work through operating procedure draft. I will supply Process facts, roles, prerequisites, approved steps, and exception rules. Produce Procedure draft; Decision/exception checkpoints; Completion evidence checklist. Separate supplied facts, assumptions and unresolved questions. Ask for missing evidence instead of inventing it.
Check the deliverables
- Procedure draft: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Decision/exception checkpoints: trace material statements or calculations to the supplied input; mark anything you cannot verify.
- Completion evidence checklist: trace material statements or calculations to the supplied input; mark anything you cannot verify.
Before you use a result
Start with fictional or appropriately redacted input, check the assistant’s file support, and review its output with the relevant analyst or data owner. A plausible answer is not proof that the task is correct. Do not authorize external actions through instructions embedded in source material.
This limited brief describes the task and review approach. The purchased pack adds its staged runbook, templates, synthetic worked example and acceptance worksheet. No AI subscription, hosted execution or runnable Python toolkit is included in an individual skill.
Buy guided edition $0.99Edition notes ↗ Review this skill

CHOOSE WHAT FITS

Three ways to start.

Launch prices in USD. One payment for the named edition. Any applicable taxes are shown at checkout.

ONE TASK

One guided skill

$0.99

One instructional pack with prompts, templates and acceptance checks. Use your own AI assistant. No runnable toolkit is included.

Find a skill

ONE AREA OF WORK

A category pack

$6.99

17-19 guided skills in your chosen category. Selected categories also include a matching Python toolkit; check the exact contents below.

Choose a category

THE FULL LIBRARY

The complete collection

$19.99

All 108 guided skills. Includes all 3 runnable toolkits as separate ZIP downloads.

Get the complete collection $19.99Edition notes ↗Review the collection ↗

Exact contents of all 6 category packs $6.99 each

Choose the category that fits your task. The matching pack below shows its existing purchase link and full contents.

Analysis19 guided skills · Guided instructions only
AI design19 guided skills + RAG & Chatbot Evaluation (Python toolkit)
Automation19 guided skills + MIS Reporting & Reconciliation (Python toolkit)
Customer intelligence17 guided skills + Feedback Intelligence (Python toolkit)
NLP & retrieval17 guided skills · Guided instructions only
Data engineering17 guided skills · Guided instructions only

3 PRACTICAL STARTING POINTS

Bring your data.
Run a real example.

Local Python tools with synthetic practice data, checked sample outputs and setup guides. Included with the matching category pack and the complete collection.

CX and product analysts

Feedback Intelligence

Turn a feedback CSV into a traceable theme report, an evidence ledger and a review queue.

BRING: A feedback CSV and your explicit tagging rules
GET: Evidence CSV · Summary JSON · Readable Markdown report

Inspect the recorded sample run

These results come from the bundled synthetic example, executed locally with Python. This preview does not run a model or upload your data.

11Accepted responses

5Flagged for review

2Conflicting rows excluded

The 15-record synthetic input yields 11 accepted responses. Five need review; two conflicting rows are excluded. Literal rules provide a traceable starting point, with human review for ambiguity.

Python 3.10+ · Standard library · No API key

Deterministic rule-based tagging. Review sentiment and ambiguous language before using the findings.

Inspect demo, commands and sample report ↗

Customer intelligence category pack17 guided skills + Feedback Intelligence toolkit

Get Customer intelligence pack $6.99Edition notes ↗Review this category ↗Ask a product question ↗

SQL, Excel and reporting analysts

MIS Reporting & Reconciliation

Reconcile orders and payments with runnable Python and SQL, surfacing discrepancies and duplicates while avoiding join multiplication.

BRING: Structured CSV exports matching the supplied schema
GET: Reconciled report · Data-quality exceptions · SQL and calculation examples

Inspect the recorded sample run

These results come from the bundled synthetic example, executed locally with Python. This preview does not run a model or upload your data.

9Orders reviewed

5Mismatches surfaced

5Source rows rejected

The synthetic exports intentionally contain discrepancies. Four of nine orders match; five need investigation. Rejected rows and superseded revisions retain source lineage.

Python 3.10+ · Standard library · No API key

A local Python/SQLite reference workflow. Adapting it to your own reporting schema requires review.

Inspect demo, commands and sample report ↗

Automation category pack19 guided skills + MIS Reporting & Reconciliation toolkit

Get Automation pack $6.99Edition notes ↗Review this category ↗Ask a product question ↗

AI developers and evaluation leads

RAG & Chatbot Evaluation

Score recorded retrieval and chatbot outputs, compare a baseline with a regression and inspect release gates.

BRING: JSONL test cases and recorded predictions
GET: Per-case evaluation · Metric summary · Release-gate report

Inspect the recorded sample run

These results come from the bundled synthetic example, executed locally with Python. This preview does not run a model or upload your data.

71.4%Recall at 3

71.4%Required citation coverage

75%Decision-label match

The deliberately regressed eight-case run fails 14 automated gates. The passing baseline is also included. These are metadata and label checks; answer correctness requires the supplied human review rubric.

Python 3.10+ · Standard library · No API key

Offline metadata checks. Semantic answer quality and factual grounding still need human assessment.

Inspect demo, commands and sample report ↗

AI design category pack19 guided skills + RAG & Chatbot Evaluation toolkit

Get AI design pack $6.99Edition notes ↗Review this category ↗Ask a product question ↗

3 runnable toolkits. 108 guided skills.

One-time purchase for the named editions. Download all 3 toolkit ZIPs alongside the guided collection. Existing individual skill packs remain instructional editions.

Get skills + 3 runnable toolkits $19.99Edition notes ↗

Tested locally on Python 3.12. Synthetic results demonstrate the supplied cases; they do not establish customer outcomes, model accuracy or production readiness. Source, fixtures and tests are included. Standalone toolkit files use MIT; guided edition terms apply to the guided packs.

THE PRACTICAL BOOKSHELF

25 practical ebooks to explore.

AI, data workflows and business analytics. Read the free samples and inspect each book’s learning project before you choose.

PDF + EPUB + companion projects · Individual one-time purchases

Browse all ebooks ↗

WHAT CHANGES IN THE GUIDED EDITION

More guidance.
Same open foundation.

3 selected starter samples are free to download under MIT. Guided edition 1.1.0 adds premium guides licensed to one human user. Use them for personal, business and client work; share your own outputs, but do not share or resell the premium guides. Included starter files and previously released MIT copies retain their MIT rights. Read the license and sharing terms ↗

Compare a selected free sample and a guided edition
Included	Free sample	Guided
Skill instructions & input template	Yes	Yes
Synthetic example & worked result	Yes	Yes
Task-specific staged prompt runbook	-	Yes
Dedicated acceptance-check worksheet	-	Yes
Starter license	MIT	MIT
Premium guide license	-	One user · Read terms

A LOOK INSIDE THE GUIDED EDITION

One prompt.
One concrete check.

Selected short excerpts from the SQL Report Review premium guide. The purchased pack includes the full staged runbook and acceptance worksheet; this preview is an excerpt, not the complete guide.

Read a real runbook and worksheet excerpt

Runbook prompt excerpt

Read SKILL.md, assets/sample-schema.md, and assets/sample-report.sql. Expand the joined rows by order/payment ID and show how O1's two payments multiply its order facts. Report the incorrect C1 values before proposing a correction.

Acceptance-check excerpt

The original C1 join has three rows: O1/P1, O1/P2, O2/P3. It reports three orders and USD 280 booked instead of two and USD 180.

Use the supplied synthetic example before adapting the workflow to your own reporting schema.

Explore SQL Report Review ↗

03 / BUILD AROUND YOUR WORK

A collection for
each area of work.

Category collections include every guided skill in that category. Choose the complete collection if your work spans several areas.

CATEGORY COLLECTION

Analysis

19 guided skills

Follow a category from focused input to checked output. Every individual pack in this category is included.

See all 19 included skills and downloads

Browse only Analysis skills ↗

Downloads after checkout

collection-analysis-guided-v1.1.0.zip · the 19 guided skill packs

Guided skills need your own compatible AI assistant and review of its output; no model subscription is included. This category contains guided instructions, not a runnable Python toolkit.

$6.99Save $11.82

Get this category $6.99Edition notes ↗Review this category ↗

CATEGORY COLLECTION

AI design

19 guided skills

Follow a category from focused input to checked output. Every individual pack in this category is included. Includes the runnable RAG & Chatbot Evaluation toolkit.

See all 19 included skills and downloads

Browse only AI design skills ↗

Downloads after checkout

collection-ai-design-guided-v1.1.0.zip · the 19 guided skill packs
rag-evaluation-toolkit-v1.0.0.zip · RAG & Chatbot Evaluation · separate ZIP

Guided skills need your own compatible AI assistant and review of its output; no model subscription is included. The separate toolkit needs Python 3.10+ with the standard library, a terminal and schema-matching input files. No API key is required. Start with its supplied synthetic example.

Inspect RAG & Chatbot Evaluation requirements and example ↗

$6.99Save $11.82

Get this category $6.99Edition notes ↗Review this category ↗

CATEGORY COLLECTION

Automation

19 guided skills

Follow a category from focused input to checked output. Every individual pack in this category is included. Includes the runnable MIS Reporting & Reconciliation toolkit.

See all 19 included skills and downloads

Browse only Automation skills ↗

Downloads after checkout

collection-automation-guided-v1.1.0.zip · the 19 guided skill packs
mis-reporting-toolkit-v1.0.0.zip · MIS Reporting & Reconciliation · separate ZIP

Inspect MIS Reporting & Reconciliation requirements and example ↗

$6.99Save $11.82

Get this category $6.99Edition notes ↗Review this category ↗

CATEGORY COLLECTION

Customer intelligence

17 guided skills

Follow a category from focused input to checked output. Every individual pack in this category is included. Includes the runnable Feedback Intelligence toolkit.

See all 17 included skills and downloads

Browse only Customer intelligence skills ↗

Downloads after checkout

collection-customer-intelligence-guided-v1.1.0.zip · the 17 guided skill packs
feedback-intelligence-toolkit-v1.0.0.zip · Feedback Intelligence · separate ZIP

Inspect Feedback Intelligence requirements and example ↗

$6.99Save $9.84

Get this category $6.99Edition notes ↗Review this category ↗

CATEGORY COLLECTION

NLP & retrieval

17 guided skills

Follow a category from focused input to checked output. Every individual pack in this category is included.

See all 17 included skills and downloads

Browse only NLP & retrieval skills ↗

Downloads after checkout

collection-nlp-retrieval-guided-v1.1.0.zip · the 17 guided skill packs

Guided skills need your own compatible AI assistant and review of its output; no model subscription is included. This category contains guided instructions, not a runnable Python toolkit.

$6.99Save $9.84

Get this category $6.99Edition notes ↗Review this category ↗

CATEGORY COLLECTION

Data engineering

17 guided skills

Follow a category from focused input to checked output. Every individual pack in this category is included.

See all 17 included skills and downloads

Browse only Data engineering skills ↗

Downloads after checkout

collection-data-engineering-guided-v1.1.0.zip · the 17 guided skill packs

Guided skills need your own compatible AI assistant and review of its output; no model subscription is included. This category contains guided instructions, not a runnable Python toolkit.

$6.99Save $9.84

Get this category $6.99Edition notes ↗Review this category ↗

FROM THE COMMUNITY

Your experience.
In your words.

Honest feedback helps the next person choose and makes these resources better.

Read reviews for

All published store reviews.

Loading reviews…

Ratings reflect published reviews. Reviews are checked for spam and personal information, regardless of star rating. A verified purchase badge is added only after an order is checked.

Share a review

Read an ebook or used a skill or toolkit? Share what you tried, what helped and what could improve.

Display name What did you try? Your reviewKeep purchase keys, payment details and confidential data out of your review.

Verify a purchase (optional)

Your purchase email and order reference are used only to check your order. They are never displayed publicly. Free sample users can leave these blank.

Purchase emailOrder reference

BEFORE YOU CHOOSE

Know what you’re getting.

Which purchases include the runnable toolkits?

Customer intelligence includes Feedback Intelligence. Automation includes MIS Reporting & Reconciliation. AI design includes RAG & Chatbot Evaluation. The complete collection includes all 3 toolkits as separate versioned ZIP downloads. Individual skill purchases include the guided instructional pack described on their cards.

What do I need to run a toolkit?

Install Python 3.10 or later, extract the toolkit ZIP, and follow its README quickstart using the supplied synthetic data. The scripts use the Python standard library and run locally without an API key or model subscription. The supplied examples were tested on Python 3.12. Your input must match the documented schema. Adapting a workflow requires review; semantic answer quality and ambiguous feedback still need human assessment.

How do I use a guided edition?

Unzip the pack, open TASK-RUNBOOK.md, and gather the named inputs. Use the instructions and relevant files with your assistant, then check its result against ACCEPTANCE-CHECKS.md. Start with the included synthetic case. An assistant that supports Agent Skills can load the SKILL.md entry point; installation and file support vary.

Do I need my own AI assistant? What software is included?

Guided skills are instructional packs: bring your own AI assistant, follow the staged prompts and review its output. Assistant compatibility and file support vary. The separate runnable Python toolkits work locally without an AI assistant, API key or model subscription; install Python 3.10 or later and follow the supplied input schema. No purchase includes an AI service, model access, a hosted integration or one-to-one support.

Can I try a sample for free?

Three selected starter samples are free under MIT. The rest of the catalog provides task previews. Guided editions include the MIT starter files plus premium runbooks, acceptance checks and a usage guide. Each purchase licenses the premium guides to one human user. Read the full terms before purchase.

Can I share a purchased skill with friends or colleagues?

You may share the MIT starter files with their notices and share or sell your own resulting work. The separate standalone toolkit files also use MIT and may be shared with their license notices. Do not share or resell the premium guides, purchase key or download access. Each additional user of the premium guides needs their own purchase. Previously distributed MIT copies, including guided edition 1.0.0, retain their MIT permissions.

Does the purchase key lock the files?

No. Keep the unique key and receipt private as proof of purchase for download assistance. These are readable offline files with no activation requirement or technical copy lock. Stock downloads do not contain a per-buyer watermark.

Is this a subscription? Are future updates included?

A guided edition uses a one-time payment for the named version. There is no recurring charge or promise of lifetime updates. The listing and receipt will identify what you purchased.

When can I buy and download the guided files?

Choose a guided edition and complete the secure checkout. Lemon Squeezy delivers the purchased files and a receipt. Selected free starter samples remain available. Download recovery, setup help and purchase questions ↗ Questions about the collection? Email Sanjaay.

Useful workflows.Results you can inspect.

Choose the task.Follow the next steps.

Feedback Analysis

A task to rehearse

Check the deliverables

Before you use a result

Data to Story

A task to rehearse

Check the deliverables

Before you use a result

AI Evaluation Planner

A task to rehearse

Check the deliverables

Before you use a result

SQL Report Review

A task to rehearse

Check the deliverables

Before you use a result

VBA Automation Planner

A task to rehearse

Check the deliverables

Before you use a result

Agent Workflow Designer

A task to rehearse

Check the deliverables

Before you use a result

Three ways to start.

One guided skill

A category pack

The complete collection

Bring your data.Run a real example.

Feedback Intelligence

MIS Reporting & Reconciliation

RAG & Chatbot Evaluation

3 runnable toolkits. 108 guided skills.

25 practical ebooks to explore.

More guidance.Same open foundation.

One prompt.One concrete check.

Runbook prompt excerpt

Acceptance-check excerpt

A collection foreach area of work.

Analysis

Downloads after checkout

AI design

Downloads after checkout

Automation

Downloads after checkout

Customer intelligence

Downloads after checkout

NLP & retrieval

Downloads after checkout

Data engineering

Downloads after checkout

Your experience.In your words.

Share a review

Know what you’re getting.

Useful workflows.
Results you can inspect.

Choose the task.
Follow the next steps.

Bring your data.
Run a real example.

More guidance.
Same open foundation.

One prompt.
One concrete check.

A collection for
each area of work.

Your experience.
In your words.