second-opinion

Runs external LLM code reviews on git diffs using Codex or Gemini CLI.

Updated Apr 5, 2026
One-click install
npx skills add https://github.com/marumo333/atrox --skill second-opinion-marumo333
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: second-opinion
Source: https://github.com/marumo333/atrox/tree/main/.claude/skills/trailofbits/plugins/second-opinion/skills/second-opinion
Command: npx skills add https://github.com/marumo333/atrox --skill second-opinion-marumo333

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Getting an independent code review from a different AI model helps catch issues that a single model might miss, but manually invoking external LLM CLIs with the right diff, prompt, and flags is tedious and error-prone. ## Core Features & Use Cases - External LLM Reviews: Shells out to OpenAI Codex CLI or Google Gemini CLI to review uncommitted changes, branch diffs, or specific commits. - Structured Findings: Codex output uses a JSON schema with prioritized findings, file:line citations, and an overall correctness verdict. - Focused Reviews: Supports security, performance, and error-handling focus areas, plus optional project conventions from CLAUDE.md or AGENTS.md. - Use Case: Before opening a PR, ask for a second opinion on your branch diff and receive parallel reviews from both Codex and Gemini, with agreements and differences summarized. ## Quick Start Ask the assistant to run a second opinion review on your uncommitted changes using both Codex and Gemini with a security focus.

Frequently Asked Questions about second-opinion

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I get a second opinion code review from another AI model?▼

Invoke the skill and choose Codex, Gemini, or both. It collects the review scope and focus interactively, generates the appropriate git diff, and runs the external CLI in headless mode to produce an independent review.

Codex CLI vs Gemini CLI for automated code review?▼

Codex exec produces structured JSON findings with priorities, confidence scores, and file:line citations using an output schema. Gemini CLI uses code-review and security extensions with piped diffs. Running both in parallel lets you compare where the models agree and differ.

Can I review a branch diff or specific commit instead of uncommitted changes?▼

Yes. The skill supports three scopes: uncommitted changes including untracked files, branch diffs against the auto-detected default branch, and a specific commit via git diff sha~1..sha.

Why does the Gemini security dependency scan get skipped?▼

The dependency scan only runs when the diff touches dependency manifest files like package.json, requirements.txt, or Cargo.toml. The scan analyzes the entire dependency tree, so it is skipped when dependencies were not modified.

What happens if Codex or Gemini CLI is not installed?▼

The skill runs the tool directly and reports the install command if it fails: npm i -g @openai/codex or npm i -g @google/gemini-cli. If both were selected, it runs only the available tool and notes the skip.