review-forge is an Agent Skill for structured, auditable code review workflows. It orchestrates multi-pass LLM review, finding synthesis, human-approved fix selection, implementation, regression testing, independent verification, and status tracking.

• Reviews local changes, branches, PRs, or commit ranges.
• Synthesizes multiple review perspectives into one fix checklist.
• Uses checkboxes as the human approval boundary for fixes.
• Fixes only approved items.
• Requires tests after fixes unless testing is blocked and documented.
• Verifies fixes independently when possible.
• Keeps process artifacts isolated under code_review/.

• review: create one model-specific review file.
• synthesize: merge review files in one feature folder into summary.md.
• fix: fix checked items, run tests, and update status.
• verify: verify fixed items, inspect or rerun tests, and update status.

When no target or base is specified:

1. If there are uncommitted or staged changes, Review Forge reviews the working tree diff.
2. If the working tree is clean, it reviews main...HEAD.
3. If main does not exist, it tries master...HEAD.
4. If no reasonable base can be inferred, it asks for one.

Explicit PRs, branches, commit ranges, or base refs always override these defaults.

Review Forge groups workflow files by feature under code_review/:

code_review/<feature>/
  codex.md
  opencode.md
  cursor.md
  summary.md
  fix-plan.md
  verify.md
  status.md

Each model review should produce one file named after the model or agent. If the same model runs multiple perspectives, use a suffix such as codex-security.md.

Command outputs are intentionally simple:

• review creates one model review file only, for example codex.md.
• synthesize creates summary.md only.
• fix creates or updates fix-plan.md and status.md.
• verify creates or updates verify.md and status.md.

Timestamped files such as STATUS_20260529.md are legacy-style outputs and should not be created unless you explicitly ask to preserve historical runs.

Add code_review/ to the target repository's .gitignore unless you intentionally want to commit review process files.

The skill instructions, template field names, and machine-readable status enums are in English. Generated reports can use report_language: auto to follow the user's prompt language, or explicit values such as en, zh-CN, or ja.

Status values remain stable English enums, while display labels may be localized.

Install from GitHub with the Skills CLI:

npx skills add vikingmute/review-forge

Or use the full GitHub URL:

npx skills add https://github.com/vikingmute/review-forge

Install globally for a specific agent:

npx skills add vikingmute/review-forge -g -a codex

Install globally for multiple agents:

npx skills add vikingmute/review-forge -g -a codex -a cursor -a opencode

Update an installed copy:

npx skills update review-forge

Update only global installs:

npx skills update review-forge -g

For local development without publishing, clone it into the cross-client Agent Skills directory:

git clone https://github.com/vikingmute/review-forge.git ~/.agents/skills/review-forge

For a project-local install, copy or clone it into the project:

mkdir -p .agents/skills
git clone https://github.com/vikingmute/review-forge.git .agents/skills/review-forge

Some clients also support their own native skill locations. For Codex, this is also valid:

cp -R review-forge ~/.codex/skills/review-forge

Then invoke it from a compatible agent with:

Use $review-forge to review this branch.

If a client does not support explicit $skill syntax, ask it to use the review-forge skill or select it from the client's skills UI.

Use $review-forge review feature: checkout-refactor model: codex.

Use $review-forge review feature: checkout-refactor model: opencode perspective: security.

Use $review-forge synthesize feature: checkout-refactor.

Use $review-forge fix feature: checkout-refactor.

Use $review-forge verify feature: checkout-refactor.