agentleFS
unlockers-aiunlockers-ai/unlockers-skills-library/skills/review-security/SKILL.md

review-security

Skill4 starsChanged 3 months agoMITView on GitHub

Review branch or pull-request changes for application-level security issues — secrets, injection, path traversal, broken authz, weak crypto, vulnerable dependencies. Use before merging changes that touch an external boundary.

23 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use review-bugbot, origin and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • 0mlkrizzz655335vSkill

    review-bugbot

    缺陷导向代码审查。用 Bugbot 子代理找 bug。

    Also in 5 projects that use this

  • 0mlkrizzz655335vSkill

    origin

    origin CLI 管理。安装、登录、更新、修复 origin.cursor.com 的 CLI。"disabled-environments:

    Also in 3 projects that use this

  • nielsmadanSkill

    review-perf

    Performance analysis for algorithmic complexity, memory leaks, N+1 queries, and render issues. Use when code feels slow, after adding loops/queries, or before scaling up.

    Also in 3 projects that use this

  • PandelisZSkill

    update-cli-config

    View and modify Cursor CLI configuration settings in ~/.cursor/cli-config.json. Use when the user wants to change CLI settings, configure permissions, switch approval mode, enable vim mode, toggle display options, configure sandbox, or manage any CLI preferences.

    Also in 4 projects that use this

  • Tsumugii24Skill

    split-to-prs

    Split current work into small reviewable PRs. Use when the user asks to split a chat, set of changes, branch, or PR.

    Also in 5 projects that use this

  • 32103210Skill

    create-rule

    Create Cursor rules for persistent AI guidance. Use when you want to create a rule, add coding standards, set up project conventions, configure file-specific patterns, create RULE.md files, or asks about .cursor/rules/ or AGENTS.md.

    Also in 7 projects that use this

  • +1 more from 0mlkrizzz655335v

The file

We split this skill into the rules it gives an agent

3 rules. 0 are common, 3 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 21 edited versions we read keep these.

  • 3

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    No validation at a trust boundary (request body, message payload, uploaded file)

    line 55
  2. Distinctive:

    No rate limiting or replay protection on a new public or auth-sensitive endpoint

    line 59
  3. Distinctive:

    Test fixtures using realistic-looking secrets that could be mistaken for real ones

    line 71

“Common” is measured: kept by at least 6 in 10 of the 21 edited versions of this file we could read. “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

Security Review Skill

Reviews changes for application-level security: what an attacker could do with this diff. Examples below are Python; the vectors are not.

Scope: application code. Infrastructure and network hardening are out of scope unless the diff touches deployment configuration.

Protocol: docs/review-protocol.md — arguments, ref resolution, exhaustiveness, severity levels, report shape. Read it first. Its § 4 Be exhaustive is mandatory; the lens steps it refers to are steps 2–4 (Find the trust boundaries, Apply the security lens, Check dependencies) below.

Process

  1. Resolve refs and gather the diff — per review-protocol.md.
  2. Find the trust boundaries this diff touches. Security findings cluster at boundaries, so classify the changes first: HTTP routes, message/queue handlers, settings and environment, subprocess and shell, filesystem paths, deserialization, token and auth handling, cryptography, logging, dependency manifests. A diff touching none of these is usually a short review — say so rather than manufacturing findings.
  3. Apply the security lens

    Critical (exploitable as-is — blocks merge):

    • Secret, credential or API key committed in source, config or a test fixture
    • Shell execution with shell=True (or equivalent) on attacker-influenced input
    • pickle.loads, unsafe YAML loading, eval, exec on untrusted data
    • Query built by string concatenation or interpolation with user input
    • Path joined from user input without containment — .. traversal, especially before a read or write
    • New endpoint or message handler with no authn/authz check where its peers have one
    • Token validation disabled, signature unverified, alg=none accepted
    • TLS verification disabled
    • Hard-coded key, reused IV, MD5/SHA1 used for security, or a non-cryptographic RNG used for tokens, passwords or IDs
    • Secret, token or full PII payload written to logs
    • CORS opened to *, or an open redirect introduced

    Major (likely exploitable — fix before merge):

    • No validation at a trust boundary (request body, message payload, uploaded file)
    • Error responses leaking stack traces, internal paths or database errors
    • Authentication present but authorization missing — any authenticated user can act on another's resources. Check object-level access, not just route-level.
    • No rate limiting or replay protection on a new public or auth-sensitive endpoint
    • Dependency added with known CVEs, or unmaintained, or pinned to a range that resolves to a vulnerable release
    • Settings field holding a secret without a secret-wrapper type
    • File written non-atomically to a path other processes read
    • Long-lived task holding secrets in memory with no rotation hook

    Minor (defence in depth):

    • User identifiers logged without a consistent redaction policy
    • HTTP client with no explicit timeout
    • Permissive file modes on write
    • Security headers missing where peer responses set them
    • Test fixtures using realistic-looking secrets that could be mistaken for real ones
  4. Check dependencies. For each modified manifest, list what was added or bumped, and flag anything known-vulnerable, abandoned, or pulled from an unexpected index. Suggest running the ecosystem's audit tool (pip-audit, npm audit, cargo audit, …).
  5. Confirm § 4 of the protocol is satisfied, then generate report → reviews/security-review-issues.md

Severity Model

SeverityIndicators
CriticalExploitable as-is: secret leak, code execution, injection, broken auth, broken crypto
MajorExploitable under realistic conditions: weak validation, missing authz, vulnerable dependency
MinorHygiene gap with no immediate exploit path

Output Format

Per review-protocol.md, with one extra field that is required on every finding:

**Vector:** how it is reached — e.g. injection / path traversal / broken authz / secret in source
**Why it matters:** the attacker scenario in one concrete sentence

"Why it matters" must name what an attacker gains. If you cannot write that sentence, the finding is speculative — leave it out.

Close with a Dependency Notes section and the standard summary.

Exclusions

  • Style and lint issues — handled elsewhere.
  • Performance concerns with no denial-of-service angle.
  • Threat-model questions about infrastructure this diff does not change.

Example Usage

review-security              # Current branch vs main
review-security feat/foo     # Specific branch
review-security feat/foo dev # Different base
review-security 3574         # PR/MR by number

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.