Back to directory

skill-file-security

A command-line installer that adds security-review skills and slash commands to AI coding assistants such as Claude, Cursor and Copilot, covering the OWASP Top 10 and CWE Top 25 categories.

Visit website (opens in a new tab)security.ai-dev-skills.com
skill-file-security

skill-file-security installs a set of security-review skills into an AI coding assistant so it can flag vulnerabilities while code is being written rather than only after the fact. An interactive mode detects the project's stack, such as Next.js, Supabase or Docker, and asks a short series of questions to install only the relevant rules; a fast mode installs everything silently, for a specific assistant or for all detected ones at once.

Once installed, slash commands run a quick scan for critical issues, a full audit that produces a score out of 100 and a report file, an apply-fixes command that asks before touching code, a status command that reads score history from a local memory file, an incident-response playbook, and a plain-language explanation of a given security rule or concept. Its stated coverage spans the OWASP Top 10 and the CWE Top 25, checking for issues such as unsafe buffer allocation, cross-site scripting, SQL injection, path traversal and missing authorization.

It is installed with npx and is written in JavaScript.

More in Agent skills

See all 581
  • Provenance gate

    github.com

    A deterministic authorization gateway that sits between an AI agent and its tools, tracing where each parameter of a side-effect call came from, and blocking the call or asking a human when the source does not check out.

  • A markdown framework of system rules that restates Isaac Asimov's Four Laws of Robotics as a set of prioritized loss calculations for large-language-model agents to follow, published as a discovery file for agents and crawlers to read.

  • Vitruvius

    github.com

    An engineering research agent that runs a discover, read, synthesize, verify and review loop over technical questions, reading sources directly rather than inferring from titles, and publishing its own scored verifier benchmark, weaknesses included.