Runtime Connector Configuration

AxonFlow supports runtime connector configuration so teams can manage governed data access without hardcoding connector behavior into application code.

Configuration Priority

Connector configuration is loaded in this order:

Priority	Source	Typical Use Case
1	Database	Enterprise runtime management
2	Config file	Self-hosted community and evaluation deployments
3	Environment variables	Fallback and simple local setups

Config File Path

The main documented entry point is:

export AXONFLOW_CONFIG_FILE=/etc/axonflow/axonflow.yaml

That is the path the agent explicitly logs and uses for MCP runtime configuration. The LLM runtime also supports an LLM-specific alternative variable, but for MCP connectors you should treat AXONFLOW_CONFIG_FILE as the primary path.

Runtime Reload Behavior

The connector runtime uses cached configuration with a default TTL of 30 seconds. In practice that means:

configuration is loaded and cached
repeated connector requests use the cached config
after cache expiry, the next request reloads configuration if it changed

This is what enables runtime updates without restarting the service in the common path.

Manual Refresh Endpoints

If the per-tenant connector registry is enabled, AxonFlow exposes refresh endpoints:

POST /api/v1/connectors/refresh
POST /api/v1/connectors/refresh/{tenant}
POST /api/v1/connectors/refresh/{tenant}/{connector}
GET /api/v1/connectors/cache/stats

These endpoints are useful for operations teams managing connector changes in active environments.

Runtime Toggle

The main runtime toggle exposed in the current agent is:

export TENANT_CONNECTOR_REGISTRY_ENABLED=false

Set that only if you need to disable the per-tenant connector registry path.

Example Unified Config

version: "1.0"

connectors:
  customer_db:
    type: postgres
    enabled: true
    display_name: "Customer Database"
    connection_url: postgres://db.example.com:5432/customers
    credentials:
      username: readonly
      password: ${DB_PASSWORD}
    options:
      max_open_conns: 10
    timeout_ms: 30000

  external_api:
    type: http
    enabled: true
    connection_url: https://api.example.com
    credentials:
      api_key: ${API_KEY}
    options:
      auth_type: api_key
      api_key_header: X-API-Key

Operational Guidance

Use config files for self-hosted repeatability and reviewability.
Use database-backed runtime config when multiple teams or operators need to manage connectors centrally.
Treat connector refresh and cache stats endpoints as operational tools, not part of the normal application flow.

Why This Matters for Production AI

Runtime configuration is not just a convenience feature. It matters because governed connector access becomes shared infrastructure once multiple AI workflows, copilots, or departments depend on the same platform.

Configuration Priority​

Config File Path​

Runtime Reload Behavior​

Manual Refresh Endpoints​

Runtime Toggle​

Example Unified Config​

Operational Guidance​

Why This Matters for Production AI​

Next Steps​