Skip to content
Conduit

skill

OWASP-Security-Skills

Claude Code plugin with two OWASP-grounded security skills: audit code, APIs, Kubernetes and LLM/agent apps against OWASP standards, and review against a 14-domain secure coding checklist.

mfkocalar/OWASP-Security-Skillscommunitylistedunreviewed

Set up

Skills are installed by placing their folder where your agent loads skills.

The download comes from github.com/mfkocalar/OWASP-Security-Skills itself, not from our server, so what you get is exactly what its author published.

Install instructions come from the README

Skills do not use an MCP config. Follow the steps in the README below.

README

synced Oct 14, 2026

OWASP Security Skills

License Version Claude Code OWASP

A Claude Code plugin with two OWASP-aligned security skills: owasp-security-audit, covering seven OWASP standards (Top 10, ASVS, MASVS, API Security, Kubernetes, LLM, and Agentic Applications — LLM and Agentic Applications are counted separately), and secure-coding-practices, a 14-domain secure coding checklist re-derived from OWASP's living sources. Together they ground security-related prompts — code reviews, auth and crypto implementation, Kubernetes manifests, LLM/agent code — in standard-aligned analysis with citations.

Install the plugin into Claude Code and any security question automatically draws on the OWASP guidance most relevant to your context. See skills/owasp-security-audit/SKILL.md and skills/secure-coding-practices/SKILL.md for the full activation triggers and workflow each skill follows, and each skill's references/ directory for the underlying standard-specific lookup docs.

What it does

  • Identifies risks in code, configuration, or design across web, API, mobile, container, and AI/LLM contexts.
  • Cites the relevant OWASP standard for the situation at hand.
  • Recommends concrete fixes with code and configuration examples.
  • Flags bypasses and edge cases attackers commonly exploit.

Coverage

Every edition and source URL below is sourced from the hardened owasp-urls.json citation files (Phases 2–3 of this project) — not re-derived here.

StandardEdition coveredScope / caveat
OWASP Top 102025 (Final) — source, retrieved 2026-07-21Web application risks: access control (incl. SSRF), misconfiguration, supply chain, cryptographic failures, injection, insecure design, auth failures, integrity failures, logging/alerting, exceptional conditions
OWASP ASVS5.0.0 current edition — source, retrieved 2026-10-04Requirement summaries use the official 5.0.0 V1-V17 chapter numbering and cite IDs as v5.0.0-X.Y.Z, each checked against OWASP's official 5.0.0 mapping file at a pinned commit; confirm IDs against the edition you audit against
OWASP MASVS2.1.0 — source, retrieved 2026-07-22Mobile app security controls (iOS / Android)
OWASP API Security Top 102023 — source, retrieved 2026-07-22API-specific risks — BOLA, broken auth, resource consumption, SSRF, and more
OWASP Kubernetes Top 102022 (stable) — source, retrieved 2026-07-22Not the 2025 draft (feedback-stage only as of retrieval, no formal release). Container/cluster risks — RBAC, secrets, workload config, network policy
OWASP LLM Top 102025 — source, retrieved 2026-07-22Prompt injection, sensitive information disclosure, supply chain, output handling, unbounded consumption, and more
OWASP Agentic Applications Top 102026 — source, retrieved 2026-07-22Agent goal hijack, tool misuse, identity/privilege abuse, memory/context poisoning, and other agent-specific risks — tracked as its own standard, distinct from the LLM Top 10 row above
Secure Coding Practices (secure-coding-practices skill)Living-source derivation, no single dated edition — Cheat Sheet Series, retrieved 2026-07-2214-domain checklist re-derived from OWASP's living sources (Developer Guide, Cheat Sheet Series, Proactive Controls). The original SCP Quick Reference Guide is archived and kept only as historical origin, not cited as a current edition

What this is NOT

  • This is guidance and reference material, not a runtime scanner. It does not execute code, run tests, or make network requests — it audits code you paste or point it at, via inspection and pattern matching.
  • The example files do not yet cover every 2025 Top 10 category. As of this release, no example demonstrates A03 Software Supply Chain Failures, A04 Insecure Design, A06 Vulnerable & Outdated Components, A08 Software or Data Integrity Failures, or A10 Mishandling of Exceptional Conditions. This table discloses that gap — it does not fill it. Authoring examples for these categories is tracked as future work.

Installation

Option 1: Claude Code plugin marketplace (recommended)

claude plugin marketplace add mfkocalar/OWASP-Security-Skills
claude plugin install owasp-security-skills@owasp-security-skills

Then confirm both skills were discovered:

claude plugin details owasp-security-skills@owasp-security-skills

This is the primary, supported install path for Claude Code.

Option 2: install.sh (alternative — symlink-based, non-plugin assistants)

For assistants without plugin/marketplace support (or for a manual, symlink-based Claude Code install), clone the repo and run the installer:

git clone https://github.com/mfkocalar/OWASP-Security-Skills.git
cd OWASP-Security-Skills
./install.sh

The installer detects your OS, symlinks the skills into the right directory, and verifies the files. To link manually instead:

# Claude (macOS)
ln -s "$PWD" ~/.claude/skills/owasp-security
# Claude (Linux)
ln -s "$PWD" ~/.local/share/claude/skills/owasp-security
# GitHub Copilot
ln -s "$PWD" ~/.copilot/skills/owasp-security

Reload or restart the assistant afterward.

Usage

Just describe what you're working on and paste the code. The skill selects the matching standard automatically.

Review this REST API endpoint for OWASP API security issues.

[paste code here]

More examples:

DomainExample prompt
Web / APIReview this code for SQL injection · Audit this endpoint for BOLA
MobileIs this iOS Keychain implementation secure?
KubernetesHarden this RBAC configuration
AI / LLMHow do I prevent prompt injection in my chatbot?
ComplianceWhat ASVS L2 requirements apply to authentication here?
Secure codingReview this code against OWASP secure coding practices

Repository structure

.claude-plugin/
  plugin.json                          Plugin identity manifest (canonical version source)
  marketplace.json                     Marketplace catalog listing
skills/
  owasp-security-audit/                Structured audit skill
    SKILL.md                           Frontmatter (name + description) + workflow
    references/                        Standard-specific lookup docs + owasp-urls.json
    scripts/quick_scan.py              Regex-based lead scanner
    assets/examples/                   9 vulnerable/secure code samples
  secure-coding-practices/             Secure Coding Practices skill
    SKILL.md
    references/                        Checklist, patterns, owasp-urls.json
    assets/examples/                   2 vulnerable/secure code samples
docs/
  SKILL-STRUCTURE.md                   Canonical skill-directory convention
scripts/
  check_version_drift.py               Version single-source-of-truth check
  lint_skill_md.py                      SKILL.md frontmatter linter
LICENSE
README.md
CONTRIBUTING.md
install.sh                             Alternative symlink-based installer

Examples

Examples live per-skill under each skill's assets/examples/ directory: 9 samples in skills/owasp-security-audit/assets/examples/ covering the OWASP Top 10 and related standards, plus 2 samples in skills/secure-coding-practices/assets/examples/ for secure coding practices. Each pairs a vulnerable pattern with a secure implementation and an explanation.

FileFocus
broken-access-control.pyMissing authorization / IDOR (A01)
security-misconfiguration.pyDebug mode, default creds, missing headers (A02)
cryptographic-failures.jsWeak hashing, plaintext storage, missing TLS (A04)
injection.jsSQL injection via string concatenation (A05)
xss.htmlReflected XSS via innerHTML (A05: Injection)
logging-monitoring-failures.pyMissing logs, secrets in logs, no alerting (A09)
api-auth-bypass.jsJWT and CORS flaws (API Security Top 10)
k8s-rbac.yamlOverly permissive RBAC, unencrypted secrets (Kubernetes Top 10)
prompt-injection.txtDirect/indirect LLM prompt injection and tool misuse (LLM Top 10 / Agentic Applications)

Paste any sample into a prompt to see the skill in action:

Review this code for security vulnerabilities according to the OWASP Top 10.

[paste example code here]

Documentation

Contributing

Contributions are welcome. See CONTRIBUTING.md.

License

Released under the MIT License.

↑↓ navigate openesc close