Pro-GenAI/Agent-Action-Guard
🛡️ Safe AI Agents through Action Classifier
Intercepts tool calls from AI agents in real-time using a lightweight neural classifier trained on the HarmActions dataset to block unsafe actions before execution. Addresses a critical gap: testing revealed 95%+ of LLMs execute harmful actions when given access to dangerous tools, often while claiming refusal. Integrates seamlessly into agent loops as a middleware layer between agents and their tool implementations.
Available on PyPI.
Stars
7
Forks
5
Language
Python
License
—
Category
Last pushed
Mar 12, 2026
Monthly downloads
962
Commits (30d)
0
Dependencies
3
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/agents/Pro-GenAI/Agent-Action-Guard"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Featured in
Related agents
microsoft/agent-governance-toolkit
AI Agent Governance Toolkit — Policy enforcement, zero-trust identity, execution sandboxing, and...
ucsandman/DashClaw
🛡️Decision infrastructure for AI agents. Intercept actions, enforce guard policies, require...
vstorm-co/pydantic-ai-middleware
Middleware layer for Pydantic AI — intercept, transform & guard agent calls with 7 lifecycle...
mattijsmoens/sovereign-shield
AI security framework: tamper-proof action auditing, prompt injection firewall, ethical...
vstorm-co/pydantic-ai-shields
Guardrail capabilities for Pydantic AI — cost tracking, prompt injection detection, PII...