When should I use perform-forge-review?

perform-forge-review is useful in the following scenarios: • Automated PR/MR Analysis: Quickly analyze code changes and identify potential issues like missing error handling or logic flaws before a human developer steps in. • Cross-Platform Review Management: Standardize code review workflows across different git forges (GitHub, GitLab, Forgejo) using a unified AI-driven interface and CLI tool. • Collaborative Feedback Loops: Generate structured review comments with clear attribution and priority levels (e.g., *Important* vs. low-priority nits) to facilitate better communication between AI and human reviewers.

When & Why to Use This Skill

This Claude skill enables AI-assisted code reviews across GitHub, GitLab, and Forgejo, streamlining the Pull Request (PR) and Merge Request (MR) evaluation process. It automates diff analysis, generates context-aware comments with priority markers, and maintains a human-in-the-loop workflow to ensure high-quality code standards and efficient developer feedback loops.

Use Cases

Automated PR/MR Analysis: Quickly analyze code changes and identify potential issues like missing error handling or logic flaws before a human developer steps in.
Cross-Platform Review Management: Standardize code review workflows across different git forges (GitHub, GitLab, Forgejo) using a unified AI-driven interface and CLI tool.
Collaborative Feedback Loops: Generate structured review comments with clear attribution and priority levels (e.g., *Important* vs. low-priority nits) to facilitate better communication between AI and human reviewers.

name	perform-forge-review
description	Create AI-assisted code reviews on GitHub, GitLab, or Forgejo. Use when asked to review a PR/MR, analyze code changes, or provide review feedback.

name

perform-forge-review

description

Create AI-assisted code reviews on GitHub, GitLab, or Forgejo. Use when asked to review a PR/MR, analyze code changes, or provide review feedback.

Perform Forge Review

Create code reviews on GitHub, GitLab, or Forgejo with human oversight.

Workflow

Use scripts/reviewtool for all operations. It requires Python 3 with no external dependencies.

1. Check out the PR

scripts/reviewtool checkout 123

This checks out the PR branch and shows the diff. For GitLab/Forgejo, set the appropriate environment variables first.

2. Review the code

Read the files, understand the changes. Use git diff and git log as needed.

3. Build comments

Start a review, then add comments:

scripts/reviewtool start --pr 123 --body "Assisted-by: OpenCode (Claude Sonnet 4)

AI-generated review. Comments prefixed with AI: are unedited."

scripts/reviewtool add --pr 123 \
  --file src/lib.rs --line 42 --match "fn process_data" \
  --body "AI: *Important*: Missing error handling"

The --match flag validates the line content to prevent misplaced comments.

4. Submit

scripts/reviewtool submit --pr 123

The review is created as pending/draft. The human reviews in the forge UI and submits when satisfied.

Comment conventions

Prefixes:

AI: — unedited AI output
@ai: — human question for AI to process
No prefix — human reviewed/edited

Priority markers:

*Important*: — must resolve before merge
(none) — normal suggestion
(low) — minor nit

Review body

Do not summarize the PR changes. The body should contain:

Attribution (required)
Concerns not tied to specific lines (optional)

Avoid positive-only inline comments — they create noise.