brainstorming
brainstorming
You MUST use this before any creative work - creating features, building components, adding functionality, modifying behavior, or planning anything new, in software or out of it (a talk, a business, a renovation).
review-changesReview changed code, blast radius, test gaps, and PR comments with code-review-graph’s delta and PR workflows.
Independent GitHub snapshot · Automated classification · Original documentation governs
Use this sub-skill when the task is to review a diff, explain risk, summarize impacted flows, or produce a GitHub PR comment from detect-changes output.
get_minimal_context_tool(task="review changes")
Read references/review-workflows.md for the exact review flow, token-efficiency rules, and when to use each tool. Read references/troubleshooting.md when detect-changes misses symbols, the token-savings panel looks wrong, or markdown escaping/relative paths need checking.
| User task | Do this |
|---|---|
| "Review my last change" | Use get_minimal_context_tool, then detect_changes_tool or get_review_context_tool depending on risk. |
| "Review this PR" | Use get_minimal_context_tool, get_review_context_tool, get_affected_flows_tool, and detect_changes_tool with the PR base. |
| "Show blast radius" | Use get_impact_radius_tool or the review context workflow. |
| "Which functions lack tests?" | Use detect_changes_tool and inspect the test-gap section. |
| "Generate PR comment" | Feed detect-changes JSON to scripts/render_pr_comment.py. |
| "Explain token savings" | Read the compact savings fields in context_savings and the render script. |
The package exposes structured review helpers for:
get_minimal_contextget_review_contextget_impact_radiusdetect_changesThe generated comment renderer escapes markdown control characters, relativizes runner paths, and keeps the hidden sticky marker as the first line so GitHub comments can be updated in place.
install-and-setup.graph-exploration.integrations-and-extensions.Native tests that ground this route include change-analysis CLI tests, context-savings unit tests, PR comment rendering tests, and split-workflow security tests. Run them only during final verification, not during ordinary skill use.