review-respond

Respond Flow (--respond mode)

When --respond is passed, this flow automates the developer's side of the review cycle. It reads the existing review, classifies each finding based on local code changes, verifies fixes are correct, runs a self-check on the fix diff, detects additional changes, and posts a structured response that the reviewer's next re-review (Step 6) can parse directly.

Respond Step 0: Initialize Session Temp Directory

Before any /tmp write, mint a per-invocation session dir so parallel /air:review --respond runs (or a respond + concurrent review in two Claude Code sessions) don't overwrite each other's wiki files, diffs, or output comments. Capture the printed path and substitute it into every $AIR_TMP reference downstream.

If the orchestrator already minted $AIR_TMP (e.g. /air:review --respond routed through review.md Step 0), reuse it — don't double-mint. Otherwise mint a fresh dir:

if [ -z "$AIR_TMP" ]; then
  find /tmp -maxdepth 1 -name 'air-*' -mtime +1 -exec rm -rf {} + 2>/dev/null
  AIR_TMP=$(mktemp -d "/tmp/air-respond-XXXXXX")
fi
echo "$AIR_TMP"

Respond Step 1: Find the review and PR

1. Detect the current branch's PR:

PR_NUMBER=$(gh pr view --json number --jq '.number' 2>/dev/null)

If no PR exists: "No open PR found on this branch. Push and create a PR first." and STOP.

1. Fetch the review comment (same query as Step 2 in the main review flow).

IMPORTANT: API responses containing comment bodies have markdown with newlines and special characters. Do NOT store the full response in a shell variable — it corrupts control characters. Pipe directly to a parser:

gh api repos/<owner>/<repo>/issues/$PR_NUMBER/comments 2>/dev/null | python3 -c "
import json, sys
comments = json.loads(sys.stdin.buffer.read())
reviews = [c for c in comments if c['body'].startswith('## Code Review')]
if reviews:
    r = reviews[-1]
    print(f'ID={r[\"id\"]}')
    print(f'CREATED={r[\"created_at\"]}')
    lines = r['body'].split('\n')
    sha = next((l.split('Reviewed at: ')[1].strip() for l in lines if 'Reviewed at:' in l), 'NOT_FOUND')
    print(f'SHA={sha}')
"

Extract:

• REVIEW_COMMENT_ID from ID= line
• REVIEW_COMMENT_CREATED from CREATED= line
• REVIEWED_AT_SHA from SHA= line
• REVIEW_COMMENT_BODY — read from the API response inside the parser, not as a shell variable