cybersecurity-skills: Instructions file for Claude Code

CLAUDE.md

Project Overview

Cybersecurity Skills is a collection of cybersecurity skill prompts for AI coding agents. Skills are authored as Claude Code SKILL.md files (the canonical format) and adapted for Cursor and Codex via a build script.

Supported Agents

Directory Structure

skills/
└── <skill-name>/
    ├── SKILL.md             # Canonical skill (35 skills)
    └── references/          # Skill-specific references + GENERATED copies of shared/ files
shared/                      # Source of truth for files several skills need (findings schema,
│                            # validators, to-sarif, severity calibration, verification, ...)
└── manifest.json            # Which skills get a copy of each file in skills/<skill>/references/
adapters/
├── cursor/                  # Generated .mdc files
└── codex/                   # Generated .md files
templates/
└── skill-template.md
scripts/
├── build-adapters.mjs       # Adapters + manifest-driven shared/ copies
├── lint-skills.mjs
├── grant-check.mjs          # allowed-tools Bash grant check (used by the lint)
└── test/                    # node:test suites + fixtures (run with npm test)
package.json                 # npm test / npm run build — no dependencies, Node >= 21
.github/
├── workflows/ci.yml   # lint, versions, plugin validate, adapter drift, tests, SkillSpector (+ weekly on main)
├── workflows/links.yml  # weekly link check; opens a broken-links issue
├── dependabot.yml     # weekly grouped bumps for the pinned actions
└── scripts/           # SkillSpector baseline splitter + SARIF merge
.skillspector-baseline.yaml
docs/
└── security-scanning.md

Skill Format (Claude Code SKILL.md)

Every skill follows the Anthropic skill creator guidelines: