agentic-security
1,848Agentic LLM vulnerability scanner and AI red-teaming kit.
DetectionPost-generationFull HITLExperimentalTier 2BEnvironment
Maturity tier details
Computed tier: 2B(raw score 4.31). Effective tier: 2B.
Facets:documented_in=0evaluated_on=0integration_surfaces=1paper_citations=0recent_commit=2stars=1.31
Source of truth:pipeline/maturity/score.py
Description
Agentic Security is a red-teaming toolkit that drives automated attack agents against LLM endpoints. Ships attack modules (multi-turn jailbreaks, role-play, data exfiltration) and scoring plugins, producing a report of identified weaknesses.
Why it's included
Agentic-attacker variant of garak: automates multi-turn attack loops rather than single-prompt probes.
Addresses failure modes
- Security vulnerability — Insecure patterns, injection flaws, weak cryptography, secret leakage.
- Scope creep — Actions outside declared task boundary.
- Fabrication — Generating reference to non-existent APIs, packages, or symbols.
Techniques
multi-turn-attack-agentsjailbreak-probesred-team-reporting