| name | context-map |
| description | Generate a focused repository context map before multi-file changes; use when you need to identify the relevant files, docs, commands, and risks. |
| category | context-docs |
| kind | atomic |
| phase | context |
| requires_write | false |
| requires_slurm | false |
| requires_benchmark_artifacts | false |
| delegates_to | [] |
| output_schema | skill_run_summary.v1 |
Context Map
When to use
Use this skill before non-trivial work when the main risk is choosing the wrong files or commands.
The objective is a compact map that reduces wrong-surface edits, not execution.
Read First
AGENTS.md
docs/dev_guide.md
docs/code_review.md
.specify/memory/constitution.md
.agents/PLANS.md for larger work
Workflow
-
Restate the task boundary
- Capture the requested outcome, likely change type, and any explicit constraints.
- Separate what is known from what still needs confirmation.
-
Identify the core files
- List the primary source files, tests, docs, and config files likely to change.
- Include nearby files that define the same contract or validation path.
-
Identify the validation path
- Name the repo-native scripts or commands that prove the change.
- Prefer
scripts/dev/ wrappers and config-first commands.
-
Identify the risk surface
- Note contracts, benchmarks, reproducibility surfaces, or docs that could drift.
- Call out any follow-up or provenance concerns.
-
Return a concise map
Primary
Adjacent
Validation
Risks
Open questions
Output
Always include the goal, relevant files/docs, likely commands, risks, and recommended next skill or
execution path.
Guardrails
- Do not edit files in this skill.
- Do not broaden scope while mapping context.
- If the task is already well-scoped, switch to the appropriate execution skill.
Proof and Guardrails
- Output should include
Primary, Adjacent, Validation, Risks, and Open questions.
- Do not include speculative recommendations beyond scoped evidence.