Agentic Security
Threat-model, audit, and review agentic AI systems against the OWASP agentic AI taxonomy.
About
Agentic Security is a Claude Code skill that encodes the OWASP Agentic AI Threats and Mitigations v1.1 taxonomy (17 threats, 6 mitigation playbooks) as an executable workflow with three routes: design, audit, and review.
Every deliverable includes a coverage matrix so no threat is skipped silently. The honest caveat: it is a single-commit, zero-star adaptation — the underlying OWASP document is authoritative, but the skill itself has no track record and OWASP does not endorse the adaptation.
What it can do
- Threat modeling (17 OWASP agentic threats)
- Security audits of agent codebases
- PR/diff security reviews
- Mitigation playbooks
- Coverage matrices
Setup
- Clone the repo: git clone https://github.com/Ramarti/agentic-security-skill.git
- Copy the skill folder: cp -R agentic-security ~/.claude/skills/ (personal) or cp -R agentic-security /path/to/repo/.claude/skills/ (project)
- Invoke with /agentic-security or ask to threat-model an agent.
Prerequisites
Claude Code. Basic familiarity with agent architectures helps.
Use cases
Works with
Trust score — 4/10
Scored 4/10 (metrics tier, metrics-v1). Evidence: 0 stars, 1 commit; OWASP v1.1 taxonomy is authoritative; skill adaptation unendorsed by OWASP; no track record. License: CC BY-SA 4.0.
★ 0 GitHub stars
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 2.0/10
- Maintenance · recency, releases, issue load
- 3.0/10
- Openness · license + docs
- 7.0/10
- Manual review · sanity read + red-flag check
- 5.0/10
- · 0 stars, 1 commit
- · OWASP v1.1 taxonomy is authoritative
- · skill adaptation unendorsed by OWASP
- · no track record
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.