Recall.ai

Fetch customer-visible runtime logs for a single Recall.ai bot in the authenticated workspace. Use this after get_bot when lifecycle events, configuration, participant activity, and chat messages are not enough to explain why a bot failed or what customer-visible errors were emitted. Requires a bot UUID. Results are returned oldest-first. Use limit to control page size and pass next_cursor back as cursor to fetch the next page. The tool only returns logs for bots in the authenticated workspace. Each returned log includes a context object with additional customer-visible details when available; use those details when they help explain the log. returned_count is the number of events in this page, not the total across all pages. If next_cursor is not null, keep passing it as cursor until next_cursor is null before treating the result as complete. Pass workspace_id to read from a different accessible workspace; omit it to use the current authorized workspace.

get_bot_logs

Fetch customer-visible runtime logs for a single Recall.ai Desktop SDK upload in the authenticated workspace. Use this when diagnosing a Desktop SDK recording and an upload UUID is available. Results are returned oldest-first. Use limit to control page size and pass next_cursor back as cursor to fetch the next page. The tool only returns logs for Desktop SDK uploads in the authenticated workspace. Each returned log includes a context object with additional customer-visible details when available. returned_count is the number of events in this page, not the total across all pages. If next_cursor is not null, keep passing it as cursor until next_cursor is null before treating the result as complete. Pass workspace_id to read from a different accessible workspace; omit it to use the current authorized workspace.

get_desktop_sdk_logs

Return current customer-visible Recall.ai service status from the incident.io customer status page for a region. Use this when the user asks whether Recall is currently reporting a service issue, outage, degradation, or incident that may affect bots, Desktop SDK, API, calendar, dashboard, or transcription. This tool only exposes status page data visible to customers; it does not expose internal incident timelines, Slack channels, roles, postmortems, or private responder notes. Component tagging is combined with title and message matching, so product_area=bots can match bot incidents even when historical component tags were incomplete. Pass workspace_id to read from a different accessible workspace; omit it to use the current authorized workspace.

get_service_status

Return current public status-page health for third-party providers used by Recall, including transcription providers and meeting platforms such as Zoom, Google Meet, Webex, and GoTo Meeting. Use this when investigating provider errors, transcription failures, upstream 429s/5xxs/timeouts, websocket session creation failures, meeting-platform incidents, or when checking whether a vendor is reporting an incident. Omit provider to check every supported provider. This tool reads public provider status pages and does not expose private vendor account, billing, or incident data. Pass workspace_id to read from a different accessible workspace; omit it to use the current authorized workspace.

get_third_party_provider_status

Get customer-visible logs for the authenticated workspace within a required timestamp window. Use this for workspace-level API, calendar connection, recording, transcript, webhook, or integration errors when get_bot_logs is too narrow. Optionally pass text to search within the bounded window. Results are oldest-first and each log includes a context object with customer-visible details when available. returned_count is the number of events in this page, not the total across all pages. If next_cursor is not null, keep passing it as cursor until next_cursor is null before treating the result as complete. Pass workspace_id to read from a different accessible workspace; omit it to use the current authorized workspace.

get_workspace_logs

List metered bot usage records for the authenticated workspace's billing account, including bot ID, platform, compute kind, recording mode, call time, and waiting-room time. Use this to answer per-bot usage questions, identify which bots ran during a period, and reconcile a customer's monthly usage breakdown. Use get_bot from the bot's workspace when additional bot metadata is needed. The time range preserves the supplied timestamps, is inclusive at start and exclusive at end, and may span at most 31 days. Results are cursor-paginated and cover every workspace on the same billing account. Use get_billed_usage for authoritative billed SKU totals, effective rates, commitments, and credits. Pass workspace_id to read from a different accessible workspace; omit it to use the current authorized workspace.

list_bot_usage

List calendar events in the authenticated workspace. Results are ordered by event start time and use cursor pagination. Pass workspace_id to read from a different accessible workspace; omit it to use the current authorized workspace.

list_calendar_events

List calendar integrations in the authenticated workspace. Results include connected and disconnected records, credential status, platform email, connection IDs that can be passed to get_calendar, and available connection timestamps. Use this when debugging calendar disconnects, removed integrations, missing calendar bots, or questions that start from a calendar email. Results use cursor pagination. Treat calendar status as diagnostic evidence; do not say you verified a calendar or workspace is connected/synced, and do not include exact calendar IDs, emails, or timestamps in user-facing answers unless the user explicitly asked for verification details. Pass workspace_id to read from a different accessible workspace; omit it to use the current authorized workspace.

list_calendars

List generated mixed-audio artifacts in the authenticated workspace, newest-first. Use this to find downloadable mixed audio for a recording or to inspect mixed-audio artifact status, format, and metadata. Results use cursor pagination; pass next_cursor back as cursor to fetch the next page. Pass workspace_id to read from a different accessible workspace; omit it to use the current authorized workspace.

list_audio_mixed

List Recall.ai documentation pages. If no category is provided, this lists pages from all categories. Without api_version, uses the selected workspace's default_docs_version. Use api_version='v1.11' for recording_config or other v1.11 features even with legacy compatibility; use v1.10 for legacy flows. Preserve explicit user versions and the version selected by search_docs. Pass either workspace_id or api_version, not both. Results include the ReadMe updated_at timestamp when available.

list_docs

List Recall.ai recordings in the authenticated workspace, newest-first. Use this when the user gives a bot ID, Desktop SDK upload ID, recording metadata, or a time range and you need to find matching recording UUIDs before inspecting artifacts or logs. Results use cursor pagination; pass next_cursor back as cursor to fetch the next page. Pass workspace_id to read from a different accessible workspace; omit it to use the current authorized workspace.

list_recordings

List the Recall.ai regions available to this MCP connection and the default region used when a tool call omits workspace_id. A single connector can span several Recall.ai regions, and each region hosts its own workspaces, bots, and recordings. Use list_workspaces to see which workspace lives in which region; to work with a workspace in another region, pass that workspace's UUID as workspace_id rather than a region name.

list_regions

List historical customer-visible Recall.ai status page incidents for a region. Use this for backtesting, support automation, or to check whether Recall publicly reported an incident during a time window. This tool reads incident.io status page incident data only, not internal incident records. Prefer get_service_status for current health checks. Pass workspace_id to read from a different accessible workspace; omit it to use the current authorized workspace.

list_service_incidents

List public historical status-page incidents for third-party providers used by Recall, including transcription providers and meeting platforms such as Zoom, Google Meet, Webex, and GoTo Meeting. Use this when debugging whether a past bot, recording, transcription, websocket, or meeting-platform failure lined up with an upstream provider incident. Prefer get_third_party_provider_status for current provider health checks. This tool reads public provider status pages and does not expose private vendor account, billing, or incident data. Pass workspace_id to read from a different accessible workspace; omit it to use the current authorized workspace.

list_third_party_provider_incidents

List dashboard webhook messages for a resource and include delivery attempts for each message. Use this to debug whether webhook events were created, whether a specific endpoint received attempts, and what HTTP response the destination returned. Defaults to bot IDs because most support investigations start from a bot. For Calendar V2 webhooks, use resource_type calendar with a calendar integration ID returned by list_calendars. Calendar V1 webhook deliveries are not available here; investigate its created bots with resource_type bot. Pass workspace_id to read from a different accessible workspace; omit it to use the current authorized workspace.

list_webhook_deliveries

List dashboard webhook endpoints configured for the authenticated workspace. Use this when investigating missing webhook deliveries or whether a destination is active and subscribed to the relevant event types. Pass workspace_id to read from a different accessible workspace; omit it to use the current authorized workspace.

list_webhook_endpoints

List configured public Recall.ai API rate limits for the authenticated workspace. Use this when investigating 429s, rate-limit warning emails, or rate-limit increase requests. The result returns the effective limit for each public endpoint and whether it comes from the default product configuration or a workspace-specific override. Pass request_path and/or request_method to narrow the result before comparing it with get_workspace_logs 429 evidence. Pass workspace_id to read from a different accessible workspace; omit it to use the current authorized workspace.

list_rate_limits

List Recall.ai workspaces available to the authenticated account. Use this whenever the user provides a workspace, organization, or account name, then choose the matching returned workspace UUID as workspace_id for bot, calendar, docs, and log tools. Do not pass a workspace name as workspace_id.

list_workspaces

Resolve a Recall.ai recording UUID or transcript artifact UUID in the authenticated workspace. Use this when a user provides a recording, transcript, or artifact ID instead of a bot ID. Returns the parent recording, associated bot ID when present, status timestamps, transcript provider, redacted provider config, and diarization settings. Can return the full canonical transcript when include_transcript is true. Download URLs are included only for generated audio/video media that is not embedded in the response. Does not return storage URIs or raw provider data. Pass workspace_id to read from a different accessible workspace; omit it to use the current authorized workspace.

get_recording_resource

Search Recall.ai docs and API reference pages. Use concise keyword queries with product terms, endpoint names, error text, or concepts. Results include matching guide and API reference pages with slugs; call get_doc with a returned slug to read the full page. If results are weak, retry with fewer words, synonyms, or related endpoint/resource names. Use this before get_doc when you do not already have an exact slug; do not invent slugs from titles or concepts. Call get_doc with a returned slug to retrieve the ReadMe updated_at timestamp when available. Without api_version, uses the selected workspace's default_docs_version. Select api_version by the user's explicit version, then the feature or known bot schema. recording_config, realtime_endpoints, recording.done, transcript.done and other v1.11-only features require api_version='v1.11' even with workspace legacy compatibility; legacy flows use v1.10. Carry this version into get_doc. Pass either workspace_id or api_version, not both. Use docs to verify product behavior. Cite doc names or links only when the user asked for docs or a link is directly useful.

search_docs

List Recall.ai bots in the authenticated workspace, newest-first. Use this to inspect recent bots or find bot UUIDs for follow-up debugging. Filter by status, status_sub_code, platform, meeting_url, participant_email, join_at_after, join_at_before, and/or metadata key/value matches. Use participant_email with both join-time bounds to find calendar-scheduled or ad-hoc bots that observed that exact email; this filter requires mcp.bot_participants.read. Continue cursor pagination until next_cursor is null, including when a page has no matches. The status and status_sub_code filters match the current bot status; use event_status and event_sub_code to find bots by historical lifecycle events. Use fields to choose which bot fields are returned. Note: the platform filter only matches bots that have already joined a call; bots still in ready/joining states are excluded when platform is set. Pass workspace_id to read from a different accessible workspace; omit it to use the current authorized workspace.