Agentic Security — LLM Vulnerability Scanner
Red-teaming kit for agentic LLM workflows: 2000+ adversarial probes, RL-crafted injections, CI-ready.
Agentic Security — LLM Vulnerability Scanner
Agentic Security is an open-source AI red-teaming toolkit that stress-tests LLM APIs and agent workflows against 2000+ adversarial prompts spanning jailbreaks, prompt injections, data exfiltration, OWASP LLM Top-10, and RL-crafted adaptive attacks.
Key features
- 2,000+ built-in attack payloads (jailbreaks, injections, OWASP LLM01-10)
- RL-based attack generator that adapts probes based on model responses
- REST API fuzzing mode — target any OpenAI-compatible endpoint
- CI integration with GitHub Actions and pass/fail threshold config
- Dashboard with per-category vulnerability heatmap
Quick start
pip install agentic_security
# Run against any OpenAI-compatible API
agentic_security --target http://localhost:8080/v1/chat/completions --threshold 0.8
npx ai-supply add agentic-security-llm-scanner
Curated mirror of the open-source Agentic Security (Apache-2.0). Get it from the source.
Compromise signals — malicious or tampered code (leaked secrets, backdoors, a dropped executable) — reduce the score, and known dependency CVEs carry a bounded penalty (they warrant review but never QUARANTINE — update the dependency to clear). Other dangerous-by-capability traits are risk surface, expected for some capabilities. Every finding is mapped to its OWASP control below.
Findings mapped to the OWASP Top 10 for LLM Applications (2025) and the OWASP Machine Learning Security Top 10. Expand any flagged control for the exact findings — compromise reduces the score; expected/risk-surface do not, except a known CVE, which carries a small bounded penalty (high/critical → Review).
The same gate an agent runs before installing (POST /api/v1/trust/agentic-security-llm-scanner/check). Click a policy:
Consume Agentic Security — LLM Vulnerability Scanner programmatically. Authenticate with an API key or session — see Authorize an agent.
# Agents: CHECK BEFORE YOU INSTALL (no auth) — score, grade, level, capability manifest
curl https://ai-supply.store/api/v1/trust/agentic-security-llm-scanner
# Gate against your org policy (returns { pass, violations })
curl -X POST https://ai-supply.store/api/v1/trust/agentic-security-llm-scanner/check \
-H "Content-Type: application/json" \
-d '{"minGrade":"B","denyPermissions":["shell"],"denyUnknownEgress":true}'
# CLI
npx ai-supply add agentic-security-llm-scanner
# REST (install → download)
curl -X POST https://ai-supply.store/api/v1/listings/agentic-security-llm-scanner/install \
-H "Authorization: Bearer $AIM_KEY"
# MCP tool
install_listing({ "slug": "agentic-security-llm-scanner" })OpenAPI spec →Curated mirror — latest upstream source. See the repository for tagged releases.