Integration Guides

Add Nyraxis observability and governance to your AI agents in minutes. Pick your framework below.

Framework integrations

• LangChain — Python, LCEL chains, LangGraph agents
• CrewAI — multi-agent workflows
• LlamaIndex — RAG pipelines and agents
• Vercel AI SDK — Next.js apps, LangChain.js

What is agent observability?

AI agents are non-deterministic, long-running, and often autonomous. Traditional APM tools weren't built for them. Nyraxis gives you three pillars:

• Traces — hierarchical records of every agent execution (LLM calls, tool calls, sub-agents)
• Metrics — cost, latency, error rate, token usage per agent and model
• Governance — policy enforcement: PII detection, prompt injection, cost limits, tool authorization

Why it matters

• Agents can leak PII in outputs without you knowing
• A single runaway agent can spend $1,000+ in minutes
• Prompt injection attacks are invisible without input scanning
• Regulators (EU AI Act) require audit trails for automated decisions

Getting started

1. Deploy Nyraxis: docker compose up
2. Create an API key at /dashboard/api-keys
3. Install the SDK: pip install nyraxis-sdk
4. Add 2 lines to your agent (see guides above)
5. Create governance policies at /dashboard/governance