Examples
Copy-paste integrations for the stacks agent teams actually use. Each example is a complete, runnable pattern against the real SDK and API.
LangChain
Trace every chain run with the callback handler, plus signed trust events for tools.
CrewAI
Give each crew member its own identity and audit chain; log task handoffs.
Claude agent
Audit a Claude tool-use loop — every tool_use block becomes a verified event.
OpenAI Agents SDK
Wrap the client for run traces and gate risky tools behind policy.
MCP tool verification
Approve the MCP servers you trust; watch unknown ones get blocked and scored.