> ## Documentation Index > Fetch the complete documentation index at: https://docs.coderabbit.ai/llms.txt > Use this file to discover all available pages before exploring further. # Get MCP server usage by pull request > Returns one JSON row per merged, reviewed pull request that used MCP, with MCP servers and tools nested under servers. Join to GET /v1/metrics/reviews on pr_url. Counts are best-effort usage telemetry. Access is controlled by an organization-scoped feature flag. Pagination is in write order rather than by merge date, so a pull request written after your cursor already passed its merge date (a backfill) still appears on a later page instead of being silently skipped. Each page request consumes one token when admitted by the request rate limiter. Reviews, review comments, and MCP share the Metrics quota for the configured quota owner. With verified owner scope, coordinate calls across API keys for the same organization, workspace, or Git-provider instance; changing an organization selector does not add quota. ## OpenAPI ````yaml /openapi.json get /v1/metrics/mcp openapi: 3.0.3 info: title: CodeRabbit API spec version: 1.0.0 servers: - url: https://api.coderabbit.ai security: - ApiKeyAuth: [] paths: /v1/metrics/mcp: get: tags: - Metrics summary: Get MCP server usage by pull request description: >- Returns one JSON row per merged, reviewed pull request that used MCP, with MCP servers and tools nested under servers. Join to GET /v1/metrics/reviews on pr_url. Counts are best-effort usage telemetry. Access is controlled by an organization-scoped feature flag. Pagination is in write order rather than by merge date, so a pull request written after your cursor already passed its merge date (a backfill) still appears on a later page instead of being silently skipped. Each page request consumes one token when admitted by the request rate limiter. Reviews, review comments, and MCP share the Metrics quota for the configured quota owner. With verified owner scope, coordinate calls across API keys for the same organization, workspace, or Git-provider instance; changing an organization selector does not add quota. operationId: mcp-metrics parameters: - in: header name: x-coderabbitai-api-key required: true schema: type: string - in: query name: org_id required: false schema: type: string description: >- Workspace API keys only; scopes the request to one provider organization. - in: query name: start_date required: true schema: type: string format: date - in: query name: end_date required: true schema: type: string format: date - in: query name: organization_ids schema: type: string description: >- Comma-separated provider organization IDs. Self-hosted only; maximum 10. - in: query name: repository_ids schema: type: string description: Comma-separated provider repository IDs; maximum 10. - in: query name: user_ids schema: type: string description: Comma-separated provider PR-author IDs; maximum 10. - in: query name: limit schema: type: integer minimum: 1 maximum: 1000 default: 1000 - in: query name: cursor schema: type: string description: Opaque next_cursor from the previous response. responses: '200': description: MCP server usage for merged pull requests content: application/json: schema: type: object required: - data - next_cursor properties: data: type: array items: type: object required: - pr_url - organization_id - repository_id - author_id - merged_at - servers properties: pr_url: type: string organization_id: type: string repository_id: type: string author_id: type: string nullable: true merged_at: type: string format: date-time servers: type: array description: MCP servers used during the review. items: type: object required: - server_name - insight_count - tools properties: server_name: type: string description: >- Most recently recorded MCP server name for this PR, or the current connection name/internal ID for historical rows without a snapshot. insight_count: type: integer description: >- Best-effort count of review insights that cited this server as a source. tools: type: array items: type: object required: - tool_name - call_count properties: tool_name: type: string description: >- Original MCP tool name when recorded; historical rows may contain the previously stored agent-facing name. call_count: type: integer description: Recorded calls for this tool. next_cursor: type: string nullable: true '400': description: >- Invalid query parameters, or the response exceeds 16 MB (RESPONSE_TOO_LARGE). Add filters or reduce the limit. '401': description: Unauthorized - invalid or missing API key '403': description: Forbidden - Enterprise plan required or agentic key not allowed '404': description: No matching organization exists '429': description: >- Request rate limit exceeded (RATE_LIMIT_EXCEEDED), active metrics query capacity reached (METRICS_BUSY), or an identical authorized query is still running (METRICS_QUERY_IN_PROGRESS). Wait at least the full Retry-After interval after receiving the response, then retry the same cursor and query parameters. Do not advance the cursor on an error. Other calls sharing the quota can cause another rejection after the wait. X-RateLimit headers are present only for RATE_LIMIT_EXCEEDED, not for the other 429 errors. headers: Retry-After: schema: type: integer description: >- Seconds to wait before retrying; dynamic for request rate limits, 5 for METRICS_BUSY or METRICS_QUERY_IN_PROGRESS X-RateLimit-Limit: required: false schema: type: integer description: >- Token bucket burst capacity, not the number of requests allowed per minute X-RateLimit-Remaining: required: false schema: type: integer description: Whole tokens remaining; 0 on request rate rejection X-RateLimit-Reset: required: false schema: type: integer description: >- Estimated Unix timestamp in seconds when the bucket will be full if no more tokens are consumed, or when a temporary local rate block ends. Use Retry-After to schedule the next retry. content: application/json: schema: type: object properties: errors: type: array items: type: object properties: message: type: string code: type: string enum: - RATE_LIMIT_EXCEEDED - METRICS_BUSY - METRICS_QUERY_IN_PROGRESS '500': description: Internal server error '503': description: >- Metrics are preparing (METRICS_PREPARING), active-query admission cannot safely proceed (METRICS_UNAVAILABLE), or API quota custody cannot be proved (POLICY_SCOPE_UNAVAILABLE). Retry the same request after Retry-After when present; otherwise retry later. headers: Retry-After: required: false schema: type: integer description: >- Seconds to wait for METRICS_PREPARING, or 5 for METRICS_UNAVAILABLE; absent for POLICY_SCOPE_UNAVAILABLE content: application/json: schema: type: object properties: errors: type: array items: type: object properties: code: type: string enum: - METRICS_PREPARING - METRICS_UNAVAILABLE - POLICY_SCOPE_UNAVAILABLE message: type: string security: - ApiKeyAuth: [] components: securitySchemes: ApiKeyAuth: type: apiKey in: header name: x-coderabbitai-api-key description: >- API key for authentication. You can create an API key from the CodeRabbit dashboard. ```` This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.