SKILL.md
Task Analyzer
Use [skills-index.yaml](references/skills-index.yaml) as the available skill catalog. Documentation routing and workflow Structural Scale belong to documentation-criteria, not this skill.
Process
1. Identify Task Essence
State the observable purpose beyond the surface operation. Preserve an explicitly invoked recipe or governing artifact as the entry point.
2. Match Skills to Task Evidence
Extract task-evidence tags and match them to the catalog. Add a skill only when its rules change the requested action, verification, or handling of a concrete risk.
| Task evidence | Consider |
|---|---|
| Observed defect or failure | ai-development-guide, testing-principles |
| Code implementation or refactoring | coding-principles, testing-principles |
| Requested design artifact | documentation-criteria |
| Multiple credible implementation strategies requiring cost comparison | implementation-approach |
| Observable cross-boundary behavior that cannot be proven more cheaply | integration-e2e-testing |
| React or TypeScript frontend code | typescript-rules and applicable frontend testing rules |
Select in this order:
governing: defines the requested output or selected workflow.risk-control: changes proof or handling of an activated failure mode.supplementary: resolves a concrete remaining risk.
3. Generate Execution Guidance
Generate only warnings and questions that can change skill selection, verification, escalation, or the first action. Prefer the smallest evidence-gathering action that can establish the target or cause.
Task analysis does not own Structural Scale, file-count estimation, documentation requirements, approval gates, implementation phases, or subagent topology.
Output
taskAnalysis:
essence: <fundamental purpose>
extractedTags: [<task evidence tag>]
selectedRules:
- skill: <skill name from skills-index.yaml>
priority: <governing|risk-control|supplementary>
reason: <how it changes execution or verification>
sections: [<relevant section name>]
metaCognitiveGuidance:
taskEssence: <fundamental purpose>
pastFailures: [<applicable known failure pattern>]
potentialPitfalls: [<task-specific risk>]
firstStep:
action: <smallest evidence-gathering or execution action>
rationale: <why it comes first>
metaCognitiveQuestions: [<question that can change the approach>]
warningPatterns:
- pattern: <applicable warning>
mitigation: <proportionate response>Return skill names and relevant section names. The consumer loads the named skills; filesystem paths, catalog metadata, and skill bodies remain at their source.
Completion Check
- Task essence, tags, and first action are tied to the current request.
- Every selected skill changes execution, verification, or a concrete risk response.
- The selected set is the smallest sufficient set.
- Questions and warnings are task-specific and proportionate.
- Structural Scale and workflow routing remain with their owning process.