code-review-gate

code-review-gate is a portable review kit for AI-assisted code review.

It provides:

• a review policy with 8 hard gates
• review artifact generation to HTML, DOCX, and JSON
• compare reporting against a baseline
• validator logic that blocks completion when hard gates are not cleared
• adapter and installer support for multiple coding-agent CLIs

Contents

• SKILL.md
• review-policy.yaml
• references/
• scripts/
• portable/install-kit.py
• adapters/

Scope

The kit is generic. It does not depend on a specific real project or repository.

Output

At review time, artifacts are generated under the current workspace:

• ./CODE_REVIEW/{YYYYMMDD}_ROUND_{seq}/

Installation

User-level install example:

npm install -g code-review-gate
code-review-gate install --agent all --force

Project-level install example:

python portable/install-kit.py --agent claude,opencode,copilot,gemini --scope project --project-root . --force

Typical Flow

1. Collect context
2. Review and update gate decisions
3. Generate compare/report artifacts
4. Validate hard gates

Example commands:

python scripts/collect_review_context.py --workspace . --author-name "Reviewer" --review-date 2026-05-12 --story-id STORY-1 --requirement-source "requirements/story-1.md" --base-ref origin/main
python scripts/build_compare_report.py --workspace . --review-dir .\CODE_REVIEW\20260512_ROUND_01
python scripts/build_review_report.py --workspace . --review-dir .\CODE_REVIEW\20260512_ROUND_01
python scripts/validate_review_gates.py --workspace . --review-dir .\CODE_REVIEW\20260512_ROUND_01

npm Publish

npm login
npm publish --access public