differential-review
An agent skill by trailofbits, from trailofbits/skills. Tags: analysis, code-quality, debugging, review, security.
What it does
Performs security-focused differential review of code changes. Adapts analysis depth to codebase size, uses git blame for context, calculates blast radius by counting callers, checks test coverage of modified code, and generates a markdown report. Use when reviewing a PR, commit, or diff for security vulnerabilities, checking whether a change re-introduces a previously fixed bug, asking what else a change could break, or finding which modified code has no test covering it.
Install
With the skills CLI, which installs into Claude Code, Codex, Cursor and other agents:
npx skills add trailofbits/skills --skill differential-reviewOr 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/differential-review/skills/differential-review ~/.claude/skills/differential-reviewSafety box score
Grade B, low risk: 25 of 100 risk points, rated 2026-10-12. 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 code-review assistant that shells out to git/grep and can query GitHub PR metadata, then writes a markdown report either in the repo or in `~/.claude/...`. The main risks are local shell execution, remote PR metadata access, and persistent file creation outside the project; it does not request secrets or privileged installs.
| Category | Level | Why |
|---|---|---|
| Code & shell execution | 2 (moderate) | The skill grants `Bash` and explicitly uses shell commands like `git checkout <baseline_commit>`, `git diff`, `git log`, `git blame`, `find`, and `grep`, so the agent can execute local commands in the repo. |
| Network access | 2 (moderate) | Phase 0 includes `gh pr view <number> --json files,additions,deletions`, which contacts GitHub and may use the user's authenticated CLI session. |
| Filesystem reach | 2 (moderate) | `reporting.md` allows report output in the current directory, the user's Desktop, or `~/.claude/skills/differential-review/output/`, so it creates persistent files outside the project. |
| Secrets & credentials | 0 (none) | The files do not ask for tokens, keys, passwords, `.env` contents, keychain items, or browser sessions; only generic GitHub CLI access is mentioned. |
| Privilege & persistence | 0 (none) | There is no `sudo`, global install, service registration, or shell/profile modification in the shipped files; state changes are limited to repo operations and report writing. |
| Instruction hijack surface | 0 (none) | The docs instruct the agent to follow the review workflow, but they do not tell it to ignore higher-priority instructions, hide actions, or follow remote content. |
| Irreversible actions | 1 (low) | The skill automatically produces a markdown report and may invoke `issue-writer`, but it does not push, deploy, pay, book, or otherwise act on external systems. |
| Transparency | 0 (none) | All shipped files are readable text or SVG; there are no binaries, base64 blobs, or downloaded installers. |
Before installing
- Confirm you are okay with the skill running shell commands like `git`, `grep`, and `find` in your checkout.
- Decide whether `gh pr view` should be allowed to contact GitHub with your logged-in CLI session.
- If you do not want hidden home-directory artifacts, change the report path away from `~/.claude/skills/differential-review/output/`.
Rated 2026-10-12 by an automated review of 7 files from the skill's repository.
Source
- Repository
- trailofbits/skills (all skills from this repository)
- Path
- plugins/differential-review/skills/differential-review/SKILL.md
- Branch
- main
- Collection
- differential-review
- Updated
- 2026-09-19
Related skills
- audit-context-building — Understand a codebase before looking for bugs in it - what each function assumes, what it guarantees, and what it depends on elsewhere.
- c-review — Performs comprehensive C/C++ security review for memory corruption, integer overflows, race conditions, and platform-specific vulnerabilities.
- constant-time-analysis — Detects timing side-channel vulnerabilities in cryptographic code.
- rust-review — Performs comprehensive Rust security review for safe/unsafe boundary issues, memory safety in unsafe blocks, concurrency hazards, panic-induced DoS.
- thermo-nuclear-code-quality-review — Run an extremely strict maintainability review for abstraction quality, giant files, and spaghetti-condition growth.
- audit-prep-assistant — Prepares codebases for security review using Trail of Bits' checklist.