Code Review Protocol
Use get_review_context to analyze changes in one call instead of manually searching for callers and test files.
Workflow
| Step | Tool | Input | Purpose |
|---|---|---|---|
| 1 | get_review_context |
files=["a.py","b.py"], query="review" |
Discover callers, test gaps, and blast radius |
| 2 | code_read |
path, offset, limit (max 50 lines) |
Read exact code ranges flagged under KEY RISK or TEST GAPS |
| 3 | logInteraction |
topic, decision="Approved/flagged..." |
Log review outcome & architectural decisions |
Rules
- One-Shot Impact: One
get_review_contextcall replaces manualcode_find+code_impact+code_trace. - Focus on Flagged Risks: Read only the line ranges reported under
KEY RISKorTEST GAPS. - Log Decision: Always record the review verdict in the planet memory via
logInteraction.