agentleFS
Sign inStart free
asmamestasmamest/Nexus-Skills/skills/security-quality/security-review/SKILL.md

security-review

Skill1 starsChanged 17 days agoMITView on GitHub

Assess application code, architecture, dependencies, configuration, and data flows for security weaknesses. Use when reviewing authentication, authorization, APIs, secrets, input handling, infrastructure, or security-sensitive changes.

We split this skill into the rules it gives an agent

20 rules. 0 are common, 20 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 167 edited versions we read keep these.

  • 20

    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:

    Consider confidentiality, integrity, availability, and privacy impacts.

    line 24
  2. Distinctive:

    Check authentication and session-management behavior.

    line 28
  3. Distinctive:

    Verify authorization on every sensitive server-side operation.

    line 29
  4. Distinctive:

    Look for insecure direct object references and cross-tenant access.

    line 30
  5. Distinctive:

    Check role enforcement, privilege escalation, and default permissions.

    line 31
  6. Distinctive:

    Check for injection, unsafe deserialization, path traversal, and command execution.

    line 36
  7. Distinctive:

    Verify that output is safely encoded for its destination.

    line 38
  8. Distinctive:

    Check error messages for sensitive information leakage.

    line 39
  9. Distinctive:

    Look for hardcoded credentials, tokens, private keys, and sensitive configuration.

    line 42
  10. Distinctive:

    Check whether secrets are logged, returned in responses, or exposed to clients.

    line 43
  11. Distinctive:

    Verify that sensitive information is accessible only to authorized components.

    line 45
  12. Distinctive:

    If a real secret is exposed, recommend revocation or rotation.

    line 46
  13. Distinctive:

    Check environment-specific configuration and debug settings.

    line 51
  14. Distinctive:

    Check for prompt injection through user inputs, documents, and tool outputs.

    line 56
  15. Distinctive:

    Verify that tools enforce authorization independently of model instructions.

    line 57
  16. Distinctive:

    Check for unintended data exposure, excessive tool permissions, and unsafe side effects.

    line 58
  17. Distinctive:

    Test whether retrieved content can influence actions outside its permitted scope.

    line 60
  18. Distinctive:

    Use safe, authorized tests only.

    line 65
  19. Distinctive:

    Do not access real user data, exploit production systems, or perform destructive actions without explicit authorization.

    line 66
  20. Distinctive:

    Avoid claiming that a system is secure simply because no findings were identified.

    line 67

“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

Objective

Identify exploitable weaknesses, explain their impact, and recommend practical mitigations based on evidence and the application's threat model.

Workflow

  1. Establish scope
    • Identify the components, changes, assets, and trust boundaries being reviewed.
    • Understand the application's intended users and deployment environment.
    • Identify sensitive data, privileged operations, and external dependencies.
    • Review relevant security requirements and repository instructions.
  2. Build a threat model
    • Identify valuable assets and possible attackers.
    • Trace untrusted inputs and data crossing trust boundaries.
    • Identify entry points, privileges, and security assumptions.
    • Consider confidentiality, integrity, availability, and privacy impacts.
    • Prioritize threats according to likelihood and potential consequence.
  3. Review identity and access
    • Check authentication and session-management behavior.
    • Verify authorization on every sensitive server-side operation.
    • Look for insecure direct object references and cross-tenant access.
    • Check role enforcement, privilege escalation, and default permissions.
    • Confirm that revocation and expiration behavior are appropriate.
  4. Review input and output handling
    • Inspect validation, normalization, encoding, and parsing.
    • Check for injection, unsafe deserialization, path traversal, and command execution.
    • Review file uploads, redirects, URL handling, and server-side request forgery risks.
    • Verify that output is safely encoded for its destination.
    • Check error messages for sensitive information leakage.
  5. Review secrets and data protection
    • Look for hardcoded credentials, tokens, private keys, and sensitive configuration.
    • Check whether secrets are logged, returned in responses, or exposed to clients.
    • Review encryption, key management, retention, and deletion requirements.
    • Verify that sensitive information is accessible only to authorized components.
    • If a real secret is exposed, recommend revocation or rotation; removing it from the latest commit alone may not be sufficient.
  6. Review dependencies and infrastructure
    • Inspect dependency versions, known vulnerabilities, and untrusted packages.
    • Review container privileges, filesystem permissions, and exposed network services.
    • Check environment-specific configuration and debug settings.
    • Examine CI/CD permissions, build provenance, and secret handling.
    • Recommend supported dependency updates and configuration hardening where justified.
  7. Review AI and agent-specific risks
    • Check for prompt injection through user inputs, documents, and tool outputs.
    • Verify that tools enforce authorization independently of model instructions.
    • Check for unintended data exposure, excessive tool permissions, and unsafe side effects.
    • Validate model outputs before passing them into privileged operations.
    • Test whether retrieved content can influence actions outside its permitted scope.
  8. Validate findings
    • Support each finding with a concrete code path, configuration, or reproducible test.
    • Distinguish confirmed vulnerabilities from plausible concerns.
    • Use safe, authorized tests only.
    • Do not access real user data, exploit production systems, or perform destructive actions without explicit authorization.
    • Avoid claiming that a system is secure simply because no findings were identified.

Finding format

For each finding, include:

  • Severity: Critical, High, Medium, Low, or Informational.
  • Location: exact file, configuration key, or affected component.
  • Evidence: the vulnerable behavior or security assumption.
  • Attack scenario: the conditions under which the weakness could be exploited.
  • Impact: the potential consequence.
  • Recommendation: a specific, actionable mitigation.
  • Verification: how the correction can be tested.

Prioritize findings using actual exploitability and impact. Do not assign severity solely because a weakness belongs to a familiar vulnerability category.

Final report

Include the scope, assumptions, findings ranked by severity, recommended mitigations, tests performed, and remaining risks. Clearly state what was not reviewed.

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.

Check your setup, free