Analyze
Objective
- Map current behavior and constraints.
- Identify problems, risks, and missing context.
When to use
- Before proposing fixes or features.
- When debugging or evaluating existing code/flows.
Inputs
- User request and acceptance criteria.
- AGENTS.md, rules, specs (if any), and relevant code/tests/logs.
- Observed symptoms or failures.
Process
- Summarize the goal and observed behavior.
- Locate relevant code/tests and note gaps or contradictions.
- Identify probable causes or risk areas.
- List what evidence or data is missing.
- Propose next actions (investigation or plan) without changing code.
Outputs
- Findings (what is known).
- Risks/unknowns (what is missing).
- Proposed next steps or plan.
Edge cases
- If information is insufficient, stop and ask for specific inputs.
- Do not change code in this skill; stay diagnostic.