receiving-code-review

name

receiving-code-review

description

Assess reviewer's suggested change (rename, refactor, move, extract) — query tea-rags imports/churn signals on affected target, measure blast radius before agreeing or pushing back. Triggers on "reviewer suggests", "should I rename", "ревьюер предлагает", "стоит ли переименовать", "code review feedback", "PR comment about", "review feedback says", "PR comment wants", "address the review", "сделай как просит ревьюер", "code review wants me to", "let's refactor this per review". NOT for vague stylistic nits without a named symbol. Wraps superpowers:receiving-code-review with tea-rags blast-radius signals.

dinopowers: receiving-code-review

Wrapper over superpowers:receiving-code-review. Agreement with review comment backed by impact data — accepting this comment affects 3 callers or 300? — not performative "yes" without verification.

Iron Rule

tea-rags impact analysis MUST run on the review-comment's target BEFORE agreeing to implement — whenever comment names concrete symbol or file.

Correct tool (semantic_search) + correct custom impact rerank (imports: 0.5, churn: 0.3, ownership: 0.2) + correct parameters (pathPattern scoping to target, metaOnly: true) + ordering (analysis BEFORE agreement) = core value.

Comment stylistic-only (typo, spacing, unclear name without structural change): skip wrapper, pass through to superpowers:receiving-code-review. Don't fabricate target.

Chaining rule: see CHAINING.md — every dinopowers:X redirects superpowers:X. NEVER bypass wrapper.

Index freshness: see FRESHNESS.md and tea-rags/rules/index-freshness.md. No background reindex hook — worktree-plan freshness explicit (clone + per-task reindex in dinopowers:executing-plans); run mcp__tea-rags__index_codebase manually to search code edited but not yet committed, BEFORE first tea-rags call.

Step 1 — Extract review target

From review comment identify:

Element	Example
Target — symbol or file the comment proposes changing	ChunkGrouper.group(), src/core/contracts/errors.ts
Change type — rename / refactor / move / extract / inline	"rename group to aggregate"
Scope — local (same file) vs cross-cutting	"rename exported symbol" = cross-cutting

Compose:

• targetPathPattern: scoping pattern (e.g. **/chunk-grouper.ts, or broader if cross-cutting)
• intent: one-sentence framing of proposed change

Comment stylistic-only: skip to Step 4 with verdict PASS-THROUGH (non-structural). Don't invent structural implications.

Step 2 — Impact analysis call

Issue ONE mcp__tea-rags__semantic_search — SAME idiom as dinopowers:writing-plans / executing-plans / verification-before-completion:

project:     <alias from list_projects — RECOMMENDED, omit path when set>
path:        <current project path — fallback when no alias is registered>
query:       <intent from Step 1>
pathPattern: <targetPathPattern>
rerank:      { custom: { imports: 0.5, churn: 0.3, ownership: 0.2 } }
limit:       10
metaOnly:    true

Do NOT substitute:

Wrong tool	Why wrong
mcp__tea-rags__hybrid_search	Custom impact rerank tied to semantic_search
Named preset ("hotspots" / "codeReview")	Named presets miss imports weight
mcp__tea-rags__find_similar on target	Finds structural analogs, not importers
grep -rn "ChunkGrouper" for usages	Misses the ranked imports overlay; noisy with comments/strings
mcp__tree-sitter__trace_impact as primary	Structural complement; this wrapper is git-first

Do NOT pass:

• metaOnly: false — verdict inputs are signals, not content
• Different weights — must match project idiom
• filter narrowing — pathPattern already scopes

Results empty: verdict UNVERIFIABLE (target not in index). Don't fabricate.

Step 3 — Compute agreement verdict

Extract from payload: imports, commitCount, bugFixRate, blameDominantAuthor (live-line owner) + recentDominantAuthor (recent committer).

Verdict ladder:

Verdict	Triggers	Action
AGREE-WITH-SCOPE	imports > 20 OR bugFixRate > 30%	Agree, but expand scope: list all N importers in response + commit to update them OR explain why a partial change is acceptable
AGREE-DIRECT	imports ≤ 5 AND bugFixRate ≤ 20%	Agree; small blast radius; single-commit change
PUSHBACK	imports > 50 (massive blast)	Respond with impact data: "renaming X affects N modules across M authors — propose alternative or narrower change first"

Compose response block:

### Impact analysis for review comment: "<comment>"

Target: <symbol or file>
- imports: <N> (rank <R> of <K> in targetPathPattern)
- churn: <N> commits, last touched <D> days ago
- live-line owner: <blameDominantAuthor> (<blameDominantAuthorPct>%)
- recent committer: <recentDominantAuthor> (<recentDominantAuthorPct>%)
- bugFixRate: <X%>

Verdict: <AGREE-WITH-SCOPE | AGREE-DIRECT | PUSHBACK>

Step 3.5 — Trace the call route the change rides (optional, preferred when the comment names a call relationship)

Step 2 gives single-symbol blast-radius number — importer count of target. When review comment names relationship between two symbols ("change how X calls Y", "this method should not reach Z", "route this through W instead"), change rides a call route, not one symbol. Promote single fanIn to actual PATH: call mcp__tea-rags__trace_path between the two named endpoints to SEE the route the change travels and which step is highest-impact chokepoint. Requires codegraph (prime shows codegraph.symbols); when off, trace_path not registered — stay with Step 2 single-symbol fanIn, note route view unavailable.

mcp__tea-rags__trace_path(
  from="<caller symbol the comment names>",   # e.g. the X in "how X calls Y"
  to="<callee symbol the comment names>",     # e.g. the Y, or the Z it "should not reach"
  rerank="blastRadius"                          # rank each hop by fanIn / import impact
)

Read the result like this:

• dangerRanking[0] — highest-impact step ON route, not necessarily either endpoint. Chokepoint the proposed change ripples through. "What does this change touch" becomes concrete hop with measured fanIn instead of one number on named symbol.
• dangerOverlay per step — carries imports / churn for that hop, so a quiet intermediate everything routes through surfaces instead of hiding between the two named endpoints.
• Empty result — NO static call path from from to to. Review comment's premise ("X reaches Y") may be structurally false — useful pushback input: change targets a route that doesn't exist as reviewer imagines. Confirm before agreeing.

Preset selection for the trace:

Situation	rerank
"What does this change ripple through" — impact-ranked route	blastRadius
"Who else owns the steps I'd touch" — who to loop into the PR	ownership

Use rerank="ownership" to see who owns each step on route — reviewer named two symbols, but change rides hops owned by others who should be looped into review. Bound search with maxDepth / maxPaths if graph deep or branchy.

Append traced route under impact block from Step 3:

**Call route the change rides (<from> → <to>), danger-ranked:**
- chokepoint: <dangerRanking[0] symbol> @ <file>:<line>
  overlay: imports <N>, churn <Y>
- also on route: <next step> @ <file>:<line>
  owner (ownership preset): <blameDominantAuthor> — loop into review

Result empty: append **Call route:** no static path <from> → <to> — the reviewer's "X reaches Y" premise may not hold; confirm before agreeing.

Danger-ranked route makes "what does this change ripple through" concrete: a chokepoint with measured fanIn and named owner is different conversation than a single import count on the symbol reviewer happened to name.

Step 3a — Tests bound to the target

When proposed change is rename / move / extract / signature change on a symbol, tests bound to current name or signature must be identified BEFORE agreeing. Invoke Skill(tea-rags:tests-as-context) with:

recipe: "tests-at-risk"
affectedFiles: [<file containing the target symbol>]   ← single-element array OK
intent: <intent from Step 1, e.g. "rename ChunkGrouper.group to aggregate">

Recipe accepts single-element affectedFiles for symbol-scoped changes — internally uses must_not relativePath any-of filter, works identically for one or many paths.

Recipe surfaces DSL leaf test chunks exercising the affected symbol. Append to impact block:

**Tests bound to current name/signature:**
- <file>:<line> — <describe-it path>
- <file>:<line> — <describe-it path>

Recipe returned SKIP: append **Tests bound:** unavailable (no DSL test chunks indexed). Recipe returned empty list: append **Tests bound:** no obvious test bindings — the change may still affect untested paths.

Bound-test list raises cost of proposed change visibly: agreeing to a rename that breaks 6 named scenarios is different conversation than one with no test bindings. Phrasing stays runner-agnostic — list scenarios, never name a runner.

Step 4 — Invoke superpowers:receiving-code-review

Invoke Skill tool with superpowers:receiving-code-review. Prepend impact block as context. Phrase handoff as:

"Before agreeing to this review, note impact signals: …… Respond with agreement tied to scope, or respectful pushback if blast radius warrants it. Technical rigor, not performative agreement.

Chaining rule reminder: when your cycle would next invoke superpowers:verification-before-completion (or any wrapped superpowers:Y), invoke dinopowers:Y instead — see the Chaining rule section above."

Let superpowers:receiving-code-review run standard rigorous-response cycle. Wrapper informs whether agreement safe and at what scope.

Red Flags — STOP and restart from Step 2

• "Comment is clear, I'll just do it" → if names symbol/file, run Step 2 anyway
• "It's just a rename" → renames propagate through imports; always analyze
• Substituted grep/git log → redo with semantic_search + custom rerank
• Named preset instead of custom weights → redo
• Agreed before Step 2 (performative yes) → revert, restart
• Pushed back before Step 2 (defensive no) → revert, restart
• Let superpowers:receiving-code-review chain into a raw superpowers:verification-before-completion without redirecting to dinopowers:verification-before-completion → intercept and invoke wrapper instead (see Chaining rule)

Common Mistakes

Mistake	Reality
Performative "good catch" agreement	Rigorous agreement requires impact data. Run Step 2.
Blanket pushback without evidence	Defensive no is as bad as performative yes. Data first.
Use rerank: "codeReview" preset because "this is code review"	codeReview preset is for finding reviewable code, not impact of a change
find_similar to assess rename blast radius	find_similar shows analogs, not callers. Use impact rerank.
Ignore imports when comment proposes a rename	Rename affects every importer. imports count = response effort
Agree then grep for callers during implementation	Wrong order — impact analysis informs whether to agree at all
Comment names "how X calls Y" but you only ran single-symbol Step 2	The change rides a route. trace_path(from=X, to=Y, rerank="blastRadius") shows the chokepoint, not just one fanIn (Step 3.5)
Loop in only the symbol's owner when the change crosses a route	The hops between X and Y have other owners. rerank="ownership" on the route surfaces who else to loop in