Pre-Landing Code Review

SkillFiles & storage

Paranoid pre-landing code review. Reads git diff, runs two-pass analysis (CRITICAL blocks PR, INFORMATIONAL goes in body), surfaces issues with file:line references. Use when reviewing code before creating a PR, after finishing implementation, or when asked to review changes.

Available today. Use it from your connected AI after setup.

Connect ahel once, and every AI you use reads what you have installed.

Then ask your AI: use the Pre-Landing Code Review skill

What this skill tells your AI

The instructions your AI receives, as published by indigoai-us/hq-core in .claude/skills/review/SKILL.md and read by ahel’s review.

Semantic code review that reads the diff and reasons about correctness, security, and edge cases. Complements /quality-gate (mechanical checks) — run both before opening the pull request.

Process

VALIDATE BRANCH → LOAD CHECKLIST → GET DIFF → TWO-PASS ANALYSIS → REPORT

Step 1: Branch Validation

Confirm you're not on main and have changes to review.

git rev-parse --abbrev-ref HEAD   # must not be main/master
git diff origin/main --stat       # must have changes

If on main or no changes: stop and tell user.

Step 2: Load Checklist

Check for repo-local override first, fall back to shared default:

  1. Detect repo root: git rev-parse --show-toplevel
  2. Check: {repo}/.claude/review-checklist.md — if exists, use it
  3. Fallback: .claude/skills/review/checklist.md (this skill's default)

If no checklist found: stop — review requires a checklist to operate.

Step 3: Retrieve Diff

git fetch origin main 2>/dev/null
git diff origin/main

Read the FULL diff before flagging anything. Issues already addressed within the diff are not issues.

Step 4: Two-Pass Analysis

Run the checklist categories against the diff:

  • Pass 1 (CRITICAL) — blocking issues. These must be resolved before PR creation.
  • Pass 2 (INFORMATIONAL) — non-blocking issues. Included in PR body for reviewer awareness.

For each finding:

  • Cite file:line precisely
  • One line describing the problem
  • One line with the fix
  • No preamble, no "looks good overall," no filler

Step 5: Report & Interact

Output format:

Pre-Landing Review: N issues (X critical, Y informational)

CRITICAL (blocking):
- [file:line] Problem description
  Fix: suggested fix

INFORMATIONAL (non-blocking):
- [file:line] Problem description
  Fix: suggested fix

If no issues: Pre-Landing Review: No issues found.

For each CRITICAL finding: present one at a time via AskUserQuestion with three options:

  • A) Fix now — apply the suggested fix
  • B) Acknowledge — proceed despite the issue (user accepts risk)
  • C) False positive — suppress in future reviews

One issue per question. Never batch. Recommend which option and why.

After all critical issues resolved: list informational findings in a single block. Suggest including them in the PR description.

Rules

  • Read-only default. No file modifications unless user explicitly approves a fix.
  • Exhaustive reporting. Surface all findings — do not quietly skip issues.
  • No premature action. Never commit, push, or create PRs.
  • Context-aware. Read the FULL diff before commenting. Don't flag issues already fixed within the diff.
  • Terse. One-line problems, one-line fixes. No summaries, no preamble.
  • Checklist-driven. Only flag categories defined in the loaded checklist.

Integration

After review completes with no unresolved critical issues:

  • Suggest running /quality-gate for mechanical checks (types, lint, tests)
  • Suggest opening the pull request (gh pr create -R {owner}/{repo} ...), then /land to take that one PR through CI and merge, or /ship to run the full review → land → deploy → smoke → monitor pipeline

Signals

GitHub stars
84
Forks
15
Last commit
Sep 2026
Hacker News mentions
20
Advanced
Catalog kind
skill
Gateway key
review-indigoai-us
Source
github.com/indigoai-us/hq-core