AI Guardrails
Input and output controls that keep AI systems safe, compliant and on-task.
Testing guardrail techniques — prompt-injection detection, PII redaction, policy engines and output validation — to understand what protects users without making assistants useless.
- Experiments
- 3
- Open Source
- 3
- Technologies
- 10
- More to Explore
- ∞
Featured Guardrail Projects
A few highlighted experiments and prototypes.
- AI AgentsFeatured
AegiSense AI Assistant
An AI-powered assistant for identity and security operations. Uses RAG, tool calling, and guardrails to answer security questions and automate workflows.
- LangGraph
- FastAPI
- PostgreSQL
- Docker
- GuardrailsFeatured
AI Guardrails Playground
Testing input/output guardrails, content filters, and policy-based controls for safer AI agents.
- Guardrails AI
- OpenAI
- Policy Engine
- Eval
- Guardrails
PII Redaction Filter
A drop-in filter that detects and redacts personal data before prompts reach the model.
- Python
- Presidio
- FastAPI
How My Guardrails Work
A typical architecture used in these experiments.
- Input(User message)
- Input Rails(Injection, PII)
- LLM(Model call)
- Output Rails(Policy, format)
- Audit Log(Traceability)
- Response(Safe answer)
Block, rewrite or escalate
All AI Guardrails
Explore all my Guardrails experiments, from simple prototypes to advanced systems.
3 experiments
An AI-powered assistant for identity and security operations. Uses RAG, tool calling, and guardrails to answer security questions and automate workflows.
- LangGraph
- FastAPI
- PostgreSQL
Testing input/output guardrails, content filters, and policy-based controls for safer AI agents.
- Guardrails AI
- OpenAI
- Policy Engine
A drop-in filter that detects and redacts personal data before prompts reach the model.
- Python
- Presidio
- FastAPI
Experiment · Learn · Build · Share
Let’s Build the Next Generation of AI Guardrails
Check out the code, try the demos, or get in touch to collaborate on exciting ideas.