Recall.ai MCP Server – Meeting Bot Observability
Recall.ai provides a managed remote MCP server for read-only access to Recall.ai workspaces, meeting bots, calendars, recordings, service status, logs, rate limits, webhooks, and developer documentation. It is useful when AI agents need to investigate meeting bot behavior, inspect calendar-driven scheduling, trace webhook delivery, or search Recall.ai product and API docs without issuing write operations.
Overview
Recall.ai provides a managed Model Context Protocol server for MCP-compatible clients. The server connects an AI agent to Recall.ai account and workspace context, with read-only tools for meeting bots, calendars, recordings, logs, service status, rate limits, webhooks, and Recall.ai documentation. The main value is operational visibility: an agent can answer questions about why a meeting bot failed, whether a calendar connection is synced, whether a webhook was delivered, or where to find the relevant Recall.ai documentation.
What the MCP server enables
The Recall.ai MCP server exposes workspace-scoped tools after authorization. Documented tools include get_info, list_workspaces, get_service_status, list_service_incidents, list_rate_limits, list_webhook_endpoints, list_webhook_deliveries, list_bots, get_bot, list_recordings, list_audio_mixed, get_recording_resource, list_calendars, get_calendar, list_calendar_events, get_calendar_event, get_bot_calendar_events, get_bot_logs, get_workspace_logs, list_docs, search_docs, and get_doc.
These tools are explicitly read-only. They let an agent inspect existing Recall.ai resources and documentation, but they should not be represented as a way to create, update, or delete bots, recordings, calendars, or webhooks.
When to use it
Use this MCP server for support, debugging, and developer-assistance workflows around Recall.ai. Practical examples include finding the bot associated with a meeting URL, checking bot status and runtime logs after a join failure, confirming whether a recording artifact or mixed audio file finished processing, tracing missing webhook deliveries, reviewing workspace rate limits after 429 responses, or searching Recall.ai docs for a guide or endpoint.
Connection and authentication
Recall.ai supports a managed remote HTTP MCP endpoint in each Recall.ai region: US East, US West, Europe, and Asia Pacific. MCP clients that support remote HTTP MCP can connect to the regional /mcp URL and discover authentication from the endpoint. For interactive connector flows such as Claude Desktop, users connect to the regional URL and complete Recall.ai OAuth. For autonomous agents, Recall.ai documents API-token authentication with an Authorization: Bearer <API_TOKEN> HTTP header.
Key considerations
Choose the MCP URL that matches the Recall.ai region of the workspace. Tools are scoped to the workspace selected during authorization; accounts with multiple workspaces should use list_workspaces and pass the appropriate workspace_id to workspace-scoped tools. Because the MCP surface is read-only, use Recall.ai's normal API or dashboard for actions that change production state. Treat API tokens as secrets, store them outside configuration files when possible, and grant access only to workspaces the agent is allowed to inspect.
Supported Transports
streamable_http
URL: https://us-east-1.recall.ai/mcp
streamable_http
URL: https://us-west-2.recall.ai/mcp
streamable_http
URL: https://eu-central-1.recall.ai/mcp
streamable_http
URL: https://ap-northeast-1.recall.ai/mcp
Frequently Asked Questions
- When should an AI agent use the Recall.ai MCP server?
- Use it when an agent needs read-only operational context from Recall.ai, such as investigating why a meeting bot failed to join, finding a recording artifact, checking calendar-driven bot scheduling, tracing webhook delivery, or searching Recall.ai documentation.
- What does this MCP server add to an AI agent's capabilities?
- It gives the agent live, authorized access to Recall.ai workspace context and documentation through documented tools for bots, recordings, calendars, logs, service status, incidents, rate limits, webhooks, and docs search.
- What can the AI agent access or manage through this server?
- The server exposes read-only tools. An agent can inspect resources such as workspaces, bots, bot logs, recordings, audio artifacts, calendars, calendar events, webhook endpoints and deliveries, service incidents, rate limits, and documentation pages, but it is not documented as a way to create or modify those resources.
- How is authentication configured for this MCP server?
- Remote HTTP MCP clients can connect to the regional Recall.ai `/mcp` endpoint and discover authentication from it. Interactive clients complete the Recall.ai OAuth flow, while autonomous agents can authenticate by sending an `Authorization: Bearer <API_TOKEN>` header using a Recall.ai API token stored as a secret.
- Which transport should be used for this MCP server?
- Use the managed remote HTTP MCP transport for the Recall.ai region that matches the workspace: US East, US West, Europe, or Asia Pacific. Recall.ai's documentation does not require running a local stdio server for this managed integration.