Skip to content

Configuration & Settings

Fine-tune the behavior and performance of your MCP Server.

Navigate to MCP Server → Configuration → Settings to manage these parameters:

SettingDefaultDescription
Enable MCP ServerMaster on/off switch for the server
Log RequestsEnable full request/response audit logging
Global Rate Limit0 (unlimited)Global requests/minute across all keys (0 = no limit)
GET Rate Limit0 (inherit global)Optional method-specific GET requests/minute
POST Rate Limit0 (inherit global)Optional method-specific POST requests/minute
Max Payload Size1,048,576 (1 MB)Maximum POST body size in bytes
Max Batch Size50Maximum JSON-RPC batch item count
Session Timeout3600s (1 hr)Idle session timeout in seconds
Max Active Sessions500Maximum simultaneously active SSE sessions
Session Queue Size100Maximum queued responses per session
Max Log Age30 daysAuto-delete logs older than this threshold
Max Log Entry Size65,536 (64 KB)Max size of a single log entry before truncation
CORS Allowed Origins(empty)Comma-separated list of allowed domain origins
Blocked Models(empty)Comma-separated extra model names to block from MCP access
Server Instructions(Default prompt)The “brain” of the server. This massive prompt is sent to the AI to teach it how to use Odoo tools and map business terms.

The MCP Server provides comprehensive traceability for all AI interactions.

  1. Go to MCP Server → Audit → Request Logs.
  2. View timestamp, API key, endpoint called, request/response data, and error messages.
  3. Filter by date range, model, or API key for investigation.
  • Logging full request/response payloads can consume significant database space.
  • Use Max Log Entry Size to truncate large payloads.
  • Use Max Log Age to ensure automatic cleanup of old data.
  • Batching: Each API call counts against rate limits; use JSON-RPC batches for related operations.
  • Filtering: Use search with specific domains and fields to minimize payload size and processing time.
  • Sessions: Monitor session counts via the /mcp/health endpoint to detect potential resource leaks in custom agents.