What you can do with Apie
See every tool call
Trace runs and sessions in the dashboard. Replay what your agent did step by step.
Read the session story
Turn session telemetry into summaries, story timelines, action items, and guardrail recommendations.
Declare boundaries
Register expected tools and capabilities. Detect when production behavior drifts from what you declared.
Guard risky actions
Start in Monitor mode to observe. Switch to Enforce mode to block deploys, secret reads, and destructive commands.
Require human approval
Pause execution when a guardrail requires approval. Resolve in the Apie dashboard.
Instrument MCP without rewriting code
Point your MCP host at the Apie proxy. Every tool call is observed and optionally guarded.
Ship telemetry reliably
Batch events, persist the queue to disk, redact secrets, and flush on shutdown.
How it fits together
Choose your path
| If you… | Start here |
|---|---|
| Have never used Apie | Connect your first agent |
| Use MCP in Cursor or Claude Desktop | MCP proxy |
| Use LangChain, OpenAI Agents, or similar | Choose how to instrument |
| Need a readable review of what a session did | Runtime intelligence |
| Need to block production deploys | Guardrails |
| Run orchestrator + worker agents | Multi-agent pipelines |
SDKs
Apie ships official SDKs for TypeScript and Python. Documentation is organized by what you want to accomplish — code examples show both languages side by side.| Language | Package | Requirements |
|---|---|---|
| TypeScript | @apie-sh/sdk | Node.js 20+, native fetch |
| Python | apie-sdk | Python 3.10+ |
Next steps
Connect your first agent
Register an agent and send your first test event in minutes.
Choose how to instrument
Pick the right integration tier for your runtime.