agentleFS
kwiggenkwiggen/claude-code-plugin/skills/gemini-reviewer/SKILL.md

gemini-reviewer

Skill1 starsChanged 4 months agoMITView on GitHub

Dual code review: gets independent reviews from both Claude and Gemini API, then synthesizes agreements, unique findings, and a verdict. Use when users say "gemini review", "dual review", or want a second-perspective code review. Requires GEMINI_API_KEY environment variable to be set.

  • Reads credentials

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 5 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    If Gemini API key is not configured or the call fails:

    line 98
  2. Distinctive:

    Do NOT block the review because of Gemini unavailability.

    line 101
  3. Distinctive:

    If Gemini produces low-quality output, note it but still include it.

    line 107
  4. Very specific:

    Use gemini-reviewer for structured dual code reviews with synthesis (agreements, unique findings, verdict).

    line 112
  5. Distinctive:

    Before presenting a dual review, verify:

    line 116
  6. Distinctive:

    If unsure about Gemini CLI flags, run:

    line 126

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Gemini Dual Code Review Skill

Before Reviewing

  1. Check Gemini availability — Run test -n "$GEMINI_API_KEY" via Bash to verify the API key is configured. If it fails, inform the user and fall back to a Claude-only review using the code-reviewer skill.
  2. Gather context — Read related files to understand existing patterns, naming conventions, and architectural decisions.
  3. Collect the diff — Use the same scope resolution as code-reviewer:
    • 1 → Uncommitted changes (git diff + git diff --cached)
    • 2 → Last local commit (git show HEAD)
    • 3 → Unpushed local commits (git diff @{upstream}..HEAD)
    • PR# → GitHub PR diff (gh pr diff <number>)

Dual Review Process

Step 1: Send to Gemini

Pipe the diff to Gemini via Bash:

echo "<diff content>" | node {pluginDir}/dist/gemini/cli.js --prompt "Review this diff for:
1. Logic bugs, incorrect assumptions, unhandled edge cases
2. Security vulnerabilities (injection, auth, data exposure)
3. Performance issues (N+1 queries, unnecessary allocations)
4. Maintainability concerns (naming, complexity, duplication)
5. Missing test coverage

Format your response as:
## Summary
Brief overview of change quality.

## Findings
- **Critical** — Must fix (security, data loss, crashes)
- **Major** — Should fix (quality/maintainability impact)
- **Minor** — Nice to fix (lower priority improvements)

Include file:line references where possible.
Be specific and actionable. Do NOT flag pre-existing issues." --system "You are a senior code reviewer. Be thorough, specific, and actionable."

Capture Gemini's full response from the JSON output's output field.

Step 2: Claude Review

Independently review the same diff using the code-reviewer skill evaluation criteria. Do NOT look at Gemini's output first — form your own assessment.

Step 3: Synthesis

Compare both reviews and present a unified report:

Output Format

Dual Code Review

Scope: [description of what was reviewed]


Claude's Review

[Claude's independent findings using code-reviewer format]


Gemini's Review

[Gemini's response, formatted cleanly]


Synthesis

Agreements

Findings both reviewers identified (highest confidence).

Unique to Claude

Findings only Claude caught.

Unique to Gemini

Findings only Gemini caught.

Verdict

Overall assessment combining both perspectives. Note where reviewers disagree and which assessment seems more accurate based on the code.

Fallback Behavior

If Gemini API key is not configured or the call fails:

  1. Inform the user: "Gemini API key not configured — running Claude-only review."
  2. Proceed with a standard code-reviewer skill review.
  3. Do NOT block the review because of Gemini unavailability.

Guidelines

  • Both reviews must be independent — do not let one influence the other.
  • Present Gemini's output faithfully; do not rewrite or suppress its findings.
  • If Gemini produces low-quality output, note it but still include it.
  • The synthesis should add value, not just repeat findings.

When to Use This Skill

Use gemini-reviewer for structured dual code reviews with synthesis (agreements, unique findings, verdict). Use gemini-advisor (/ask-gemini) for general second opinions on any topic, file, or question — not limited to code review.

Pre-Delivery Checklist

Before presenting a dual review, verify:

  • Claude's review was completed independently before reading Gemini's
  • Gemini's output is presented faithfully without rewriting
  • Synthesis contains all 3 sections: Agreements, Unique to Claude, Unique to Gemini
  • Verdict addresses any disagreements between reviewers
  • Fallback behavior was followed if Gemini was unavailable

CLI Reference

If unsure about Gemini CLI flags, run:

node {pluginDir}/dist/gemini/cli.js --help

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.