second-opinion
An agent skill by trailofbits, from trailofbits/skills. Tags: analysis, code-review, developer-tools, github.
What it does
Gets independent code reviews from Codex or Antigravity for uncommitted changes, branch diffs, and commits. Use when the user requests an external review, a second opinion on code, a codex review, a gemini review, an antigravity review, or /second-opinion.
Install
With the skills CLI, which installs into Claude Code, Codex, Cursor and other agents:
npx skills add trailofbits/skills --skill second-opinion
Or copy the skill folder into Claude Code's skills directory by hand (~/.claude/skills for every project, or .claude/skills inside one):
git clone --depth 1 https://github.com/trailofbits/skills
cp -r skills/plugins/second-opinion/skills/second-opinion ~/.claude/skills/second-opinion
Safety box score
Grade C, moderate risk: 40 of 100 risk points, rated 2026-10-20. The grade rates what the skill and its scripts can reach on the machine of whoever installs it, across eight categories from shell execution to secrets access. How the score works.
This skill is a read-only wrapper around external code-review CLIs. Its main risks are that it captures the selected diff, including untracked files, and sends that material off-machine to third-party review services, while also shelling out to local binaries from the user’s PATH.
| Category | Level | Why |
|---|---|---|
| Code & shell execution | 2 (moderate) | It runs shell commands and external CLIs (`git`, `mktemp`, `codex exec`, `agy`, `gemini`) to prepare and submit a review, though it does not execute arbitrary project code. |
| Network access | 3 (high) | The review workflow sends the captured patch and context to external services via `codex exec`, `agy`, or `gemini`, so selected repository content leaves the machine. |
| Filesystem reach | 1 (low) | It reads the repository diff and writes temporary prompt/output/diagnostic files with `mktemp` outside the checkout, but it does not edit or delete project files. |
| Secrets & credentials | 2 (moderate) | Untracked files are explicitly included in the captured patch, and that patch is transmitted verbatim to external reviewers, so any secrets present in the reviewed changes can be exposed. |
| Privilege & persistence | 1 (low) | The references mention user-level setup such as `npm i -g ...` and checking `~/.local/bin`, but the skill itself does not request sudo, install services, or change OS settings. |
| Instruction hijack surface | 0 (none) | The skill does not tell the agent to ignore its rules, follow fetched instructions, or hide behavior; it instead says repository material must be treated as data, not instructions. |
| Irreversible actions | 1 (low) | It automatically launches external reviews and can choose a provider/model, but it does not publish, merge, deploy, pay, or modify files. |
| Transparency | 0 (none) | All shipped files are readable markdown, YAML, JSON, and SVG; there are no binaries, minified blobs, or hidden runtime-decoded payloads. |
Before installing
- Read `references/review-input.md` first; it includes untracked files in the patch and sends the whole prompt to an external CLI.
- Verify there are no secrets in the selected diff or untracked files before running the skill.
- Check that `codex`, `agy`, or `gemini` resolves to the intended binary on your PATH.
- If you care about privacy, use a scratch clone or scrub sensitive files before asking for a second opinion.
Rated 2026-10-20 by an automated review of 8 files from the skill's repository.
Source
- Repository
- trailofbits/skills (all skills from this repository)
- Path
- plugins/second-opinion/skills/second-opinion/SKILL.md
- Branch
- main
- Collection
- second-opinion
- Updated
- 2026-09-19
Related skills
- github-triage — Triages a repository's open GitHub issues and pull requests via the gh CLI.
- competitors-analysis — Discover, clone, update, and analyze competitor repositories with evidence-based competitive intelligence.
- c-review — Performs comprehensive C/C++ security review for memory corruption, integer overflows, race conditions, and platform-specific vulnerabilities.
- dwarf-expert — Analyzes DWARF debug information in compiled binaries.
- pr-improver — Runs an autonomous review-and-fix improvement loop over the current branch's changes until a PR review comes back clean.
- rust-review — Performs comprehensive Rust security review for safe/unsafe boundary issues, memory safety in unsafe blocks, concurrency hazards, panic-induced DoS.