Agent integration

Enclavia is an egress proxy. Your agent keeps its existing logic and SDKs; you only point its outbound HTTP at the data plane.

Python (recommended)

The enclavia-agent wheel patches requests and httpx at startup, so Anthropic/OpenAI SDKs and plain HTTP clients are routed through the enclave with no code changes.

pip install enclavia-agent

export ENCLAVIA_API_URL="http://<data-plane>:3001/api/agent/egress"
export ENCLAVIA_API_KEY="<workload-api-key>"
export ENCLAVIA_AGENT_ID="<agent-id>"

That's it. The patch is a no-op unless the environment variables are set, so it is safe to install everywhere.

Other languages

Requests from agents in other languages are routed through the Enclavia data plane automatically — configure the data plane endpoint and API key via environment variables or your agent's HTTP client configuration. The data plane is set up during deployment and managed through the dashboard; there is no raw HTTP API you need to call from application code.

Supported traffic