agentleFS
Sign inStart free
open-edge-platformopen-edge-platform/edge-ai-libraries/.github/skills/security-review/SKILL.md

security-review

Skill177 starsChanged 7 months agoApache-2.0View on GitHub

On-demand security review for code, containers, and Kubernetes configuration. Use when changes involve authentication/authorization logic, input parsing, file handling, secrets, logging, dependency upgrades, Dockerfiles, Docker Compose, Helm charts, CI/CD workflows, or privilege elevation. Covers secure code review (input validation, injection, dynamic execution, dependency hygiene), AI-generated code guardrails, container artifact hardening (Dockerfile, Compose), and Helm/Kubernetes security defaults. Produces classified findings (Fix in artifact vs. Deployment-time responsibility) with severity and confidence levels. Do NOT use for: runtime, host, cluster, or organizational security controls; general code quality or style reviews unrelated to security; enforcing network policies, node hardening, or cloud IAM configuration.

  • Pipes a download into a shell

We split this skill into the rules it gives an agent

29 rules. 0 are common, 24 are distinctive, and 5 are very specific to this project.

  • 0

    Common

    Most of the 167 edited versions we read keep these.

  • 24

    Distinctive

    This author's own choices. Worth a look.

  • 5

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    When reviewing AI-generated changes:

    line 37
  2. Distinctive:

    Treat AI output as untrusted draft code until reviewed and tested

    line 39
  3. Distinctive:

    Verify package names, APIs, images, and tools exist and originate from trusted sources

    line 40
  4. Very specific:

    Never accept generated code or configs that inject secrets via source files, Dockerfile ARG/ENV, or committed templates

    line 43
  5. Distinctive:

    Do not propagate unvalidated input across trust boundaries

    line 58
  6. Distinctive:

    Avoid command, query, or expression construction via string concatenation

    line 59
  7. Distinctive:

    Use parameterized queries for all database access

    line 60
  8. Distinctive:

    Do not rely on client-side enforcement for access control

    line 65
  9. Distinctive:

    Avoid ignored return values or silent failures

    line 70
  10. Distinctive:

    Avoid unchecked allocations and unbounded resource use

    line 74
  11. Distinctive:

    Ensure files, sockets, and handles are closed deterministically

    line 75
  12. Distinctive:

    Do not log credentials, tokens, secrets, or PII

    line 79
  13. Distinctive:

    Avoid shelling out when native APIs or libraries exist

    line 88
  14. Distinctive:

    Prefer latest stable versions.

    line 90
  15. Distinctive:

    If uncertainty exists, flag it clearly rather than guessing or assuming safety.

    line 97
  16. Very specific:

    Avoid latest or floating tags.

    line 111
  17. Distinctive:

    Prefer minimal base images

    line 112
  18. Distinctive:

    Ensure containers do not run as root

    line 113
  19. Distinctive:

    Avoid setuid or setgid binaries

    line 114
  20. Distinctive:

    Use multi-stage builds and remove build tools, package caches, and temp files from final image

    line 115
  21. Very specific:

    Prefer COPY over ADD

    line 116
  22. Very specific:

    Never embed secrets in ARG, ENV, or filesystem layers

    line 117
  23. Very specific:

    Avoid privileged: true and host networking unless explicitly justified

    line 121
  24. Distinctive:

    Do not mount the Docker socket

    line 122
  25. Distinctive:

    Prefer read-only root filesystem where feasible

    line 135
  26. Distinctive:

    Do not template secrets directly into charts

    line 137
  27. Distinctive:

    Document required runtime security assumptions

    line 138
  28. Distinctive:

    Do not enforce cluster-wide, node-level, or runtime security controls.

    line 140
  29. Distinctive:

    Use severity levels: Critical / High / Medium / Low with confidence: High / Medium / Low

    line 151

“Common” is measured: kept by at least 6 in 10 of the 167 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

Purpose

This document defines on-demand security review guidance for code and configuration changes. It complements the always-on secure-by-default rules defined in .github/copilot-instructions.md.

This skill applies only at development and authoring time. Runtime, host, cluster, or organizational security controls are explicitly out of scope.

Security Review Trigger Points

Load this security review skill when changes involve:

  • Authentication, session, or token logic
  • Authorization or resource ownership checks
  • Input parsing, validation, normalization, or canonicalization
  • File handling, deserialization, template rendering, or process execution
  • Logging, telemetry, secrets handling, or sensitive data paths
  • Dependency upgrades, lockfile changes, or CVE-related updates
  • Dockerfile or container base image changes
  • Docker Compose, Helm charts, or Kubernetes-related configuration
  • CI/CD workflow changes affecting build, test, release, or scanning
  • Privilege elevation, root execution, host mounts, or new Linux capabilities

AI-Generated Code Guardrails

When reviewing AI-generated changes:

  • Treat AI output as untrusted draft code until reviewed and tested
  • Verify package names, APIs, images, and tools exist and originate from trusted sources
  • Reject suggestions that bypass or disable security controls for convenience
  • Require pinned versions and lockfiles for generated dependencies; prefer integrity-verified installs when supported
  • Never accept generated code or configs that inject secrets via source files, Dockerfile ARG/ENV, or committed templates
  • Reject generated install scripts that use unchecked remote execution patterns (e.g., curl | sh) without checksum or signature verification
  • Reject generated build commands that disable TLS, certificate verification, or security checks to make builds pass
  • Apply RCI pattern: ask the AI to review its own output for security issues, then improve; repeat 1-2 iterations

Secure Code Review (OSS Context)

Apply when reviewing application logic, services, APIs, or libraries.

Input handling

  • Validate input at trust boundaries (format, type, range, length)
  • Avoid unsafe deserialization
  • Do not propagate unvalidated input across trust boundaries
  • Avoid command, query, or expression construction via string concatenation
  • Use parameterized queries for all database access

Authorization

  • Keep authorization checks server-side and close to protected actions or resources
  • Do not rely on client-side enforcement for access control

Error handling

  • Errors must not expose sensitive internal details
  • Avoid ignored return values or silent failures

Memory & resource safety (where applicable)

  • Avoid unchecked allocations and unbounded resource use
  • Ensure files, sockets, and handles are closed deterministically

Logging & telemetry

  • Do not log credentials, tokens, secrets, or PII
  • Logs should be actionable without exposing sensitive data

Dynamic execution

  • Avoid unsafe dynamic execution patterns (eval, exec, reflection, or untrusted code execution).

Dependency usage

  • Avoid shelling out when native APIs or libraries exist
  • Flag outdated, unmaintained, or suspicious dependencies
  • Prefer latest stable versions; specify exact or range-locked versions

OSS-specific review checks

  • Is externally observable security-relevant behavior documented?
  • Are assumptions and limitations stated explicitly for users?

If uncertainty exists, flag it clearly rather than guessing or assuming safety.

Container Artifact Review (Development-Time)

Apply when generating or reviewing:

  • Dockerfiles / Containerfiles
  • docker-compose.yml
  • Helm charts (templates and values)

Dockerfile

  • Avoid latest or floating tags; pin versions or digests
  • Prefer minimal base images
  • Ensure containers do not run as root
  • Avoid setuid or setgid binaries
  • Use multi-stage builds and remove build tools, package caches, and temp files from final image
  • Prefer COPY over ADD
  • Never embed secrets in ARG, ENV, or filesystem layers

Docker Compose

  • Avoid privileged: true and host networking unless explicitly justified
  • Do not mount the Docker socket
  • Restrict host filesystem mounts
  • Limit exposed ports and networks; prefer internal networks

Concerns that depend on deployment or runtime policy should be flagged as: "Deployment-time responsibility."

Helm / Kubernetes Review (Development-Time)

  • Default to runAsNonRoot: true
  • Set allowPrivilegeEscalation: false
  • Prefer read-only root filesystem where feasible
  • Drop unnecessary Linux capabilities
  • Do not template secrets directly into charts
  • Document required runtime security assumptions

Do not enforce cluster-wide, node-level, or runtime security controls.

Review Output Expectations

  • Identify which section applies (Code / Container / Helm)
  • Classify findings as:
    • Fix in artifact
    • Deployment/runtime responsibility
  • Explicitly state assumptions or uncertainty
  • Use severity levels: Critical / High / Medium / Low with confidence: High / Medium / Low
  • Include specific file/function references and recommended fixes

Security review is advisory; final decisions belong to maintainers.

Download now (ZIP)

  • Other versions192 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 192

    1. affaan-m/ECC★ 246ksame name
    2. github/awesome-copilot★ 41ksame name
    3. langfuse/langfuse★ 37ksame name
    4. nearai/ironclaw★ 13ksame name
    5. joe-bell/cva★ 6.9ksame name
    6. doorkeeper-gem/doorkeeper★ 5.5ksame name
    7. compozy/compozy★ 2.8ksame name
    8. getsentry/sentry-python★ 2.2ksame name
    9. trycompai/comp★ 2.1ksame name
    10. xu-xiang/everything-claude-code-zh★ 1.9ksame name

    182 more not listed

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • Used alongsideProjects with it also use dependency-cve-sweep, fleet-standard-adoption 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.
    • berntpoppSkill

      dependency-cve-sweep

      Use when handling Dependabot PRs, CodeQL/Trivy alerts, or doing a routine dependency/CVE sweep for this repo.

      Also in 12 projects that use this

    • berntpoppSkill

      fleet-standard-adoption

      Use when bringing this repo into compliance with a GeneFoundry fleet standard (tool-naming, response-envelope, container-hardening, versioning, MCP transport, or logging) or closing its tracking issue.

      Also in 12 projects that use this

    • niloykumarbarmanSkill

      agent-sort

      Build an evidence-backed ECC install plan for a specific repo by sorting skills, commands, rules, hooks, and extras into DAILY vs LIBRARY buckets using parallel repo-aware review passes. Use when ECC should be trimmed to what a project actually needs instead of loading the full bundle.

      Also in 12 projects that use this

    • ysyecustSkill

      brand-discovery

      Use when a brand needs to discover or articulate its identity through structured multi-session interviews. Covers purpose, positioning, audience, personality, voice, narrative, and founder-brand tension across 8 modules using laddering, 5 Whys, and projective techniques. Produces a resumable session with disk-persisted state and a master brandbook (90_SYNTHESIS.md).

      Also in 5 projects that use this

    • InfopibeSkill

      everything-claude-code

      Also in 11 projects that use this

    • royaluniondesign-sysSkill

      configure-ecc

      Interactive installer for Everything Claude Code — guides users through selecting and installing skills and rules to user-level or project-level directories, verifies paths, and optionally optimizes installed files.

      Also in 10 projects that use this

    • +2 more from berntpopp
  • In this repository3 other files it gives its agents.

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.

Check your setup, free