Skip to content

Admin Dashboard: Settings

The Settings page controls engine-wide preferences that affect all agents and API requests. Currently, it covers BYOK (Bring Your Own Key) configuration and logging levels.

BYOK (Bring Your Own Key)

BYOK allows API consumers to override the model for a single request by passing their own API key in request headers. This is useful for multi-tenant deployments where each customer uses their own LLM account.

BYOK is configured per-provider: you can enable it for OpenAI but disable for Anthropic.
When enabled, the consumer passes X-Model-Provider, X-Model-API-Key, and X-Model-Name headers.
The user-provided key is used for that single request only and is never stored or logged.
If the headers are not present, the agent uses its configured model as normal.

# BYOK headers in a request (when enabled for the provider)
curl -N http://localhost:8443/api/v1/schemas/{schema_id}/chat \
  -H "Authorization: Bearer bb_your_token" \
  -H "X-Model-Provider: anthropic" \
  -H "X-Model-API-Key: sk-ant-customer-key" \
  -H "X-Model-Name: claude-sonnet-4-20250514" \
  -H "Content-Type: application/json" \
  -d '{"message": "Hello"}'

Logging level

Change the engine’s logging verbosity at runtime without restarting:

Level	Description
`debug`	Most verbose. Logs every LLM call, tool execution, and internal state change.
`info`	Default. Logs agent activity, task lifecycle, and MCP connections.
`warn`	Only warnings and errors. Good for production with stable agents.
`error`	Only errors. Minimal output, useful for high-traffic deployments.

What’s next