FareedKhan-dev/agentic-guardrails
Layered guardrails to make agentic AI safer and more reliable.
Implements a three-layer defense-in-depth architecture using LangGraph's ReAct orchestrator, with asynchronous input validation (PII/MNPI detection, threat filtering), AI-powered action plan enforcement with human escalation, and output verification for hallucinations and compliance. Provides red-teaming capabilities and adaptive guardrails that learn from failures, integrating with OpenAI-compatible APIs and SEC EDGAR for financial agent use cases.
No commits in the last 6 months.
Stars
34
Forks
12
Language
Jupyter Notebook
License
MIT
Category
Last pushed
Oct 05, 2025
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/rag/FareedKhan-dev/agentic-guardrails"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
neural-maze/philoagents-course
When Philosophy meets AI
techwithaijack-beep/enterprise-ai-agent-blueprint
Production-ready reference architecture, checklists, and templates for enterprise AI agent systems
benitomartin/crewai-rag-langchain-qdrant
Collaborative Multi-Agent RAG with CrewAI
alexnodeland/crewlit
🚣 Crewlit brings the power of CrewAI to your browser, making multi-agent AI systems accessible...
VIVPM/multi-crew-leads-dashboard
A Streamlit-based leads pipeline app that scores leads, enriches data with web search, and...