Discover the workspace and project ids this connection may use before calling private tools. Returns accessible workspaces with their projects; it does not list other users' workspaces. Pass workspace/project ids to private tools to select a scope, or omit them to use this connection's defaults.

Create a workspace owned by the signed-in user, including its first project. Use create_project instead to organize an existing workspace. Requires an OAuth grant for all workspaces; a workspace-scoped API key cannot create one. Returns the new workspace record and id; use list_workspaces to discover its project.

Create a project inside the selected workspace and return its record and id. Use this to separate monitors within an existing workspace; use create_workspace for a separate team. Requires access to all projects in that workspace. The new project is immediately usable through the project argument.

Read monitor summaries for the selected project, including ids, current status, uptime and latency. Use this to find a monitor before changing it or to scan for outages; use get_monitor for recent check history. Secret HTTP header values are masked unless the caller may manage the workspace.

Read one monitor's configuration, recent check history and uptime statistics by its UUID from list_monitors. Use this to investigate a specific monitor; list_monitors is the summary view and run_check performs a new check. Returns the monitor plus history; secret HTTP headers are masked for callers without management access.

Create and start a monitor in the selected project; returns its record and UUID (heartbeat monitors include a ping token). Use list_regions and list_channels to discover optional region/channel ids. Heartbeats wait for pings; other types schedule outgoing checks. Auto-enabled alert channels are added, and failures can open incidents and notify channels. Hosted plans may reject capacity, frequency or region choices; private network targets are refused by default.

Patch an existing monitor and return its updated record. Read get_monitor first, then send id and at least one changed field; omitted fields stay unchanged and type cannot change. Lists and headers replace their previous values, while documented nulls clear optional settings. Prefer pause_monitor/resume_monitor for a pause-only change; use this for configuration changes or a combined edit. Resuming sets status to pending; heartbeat interval/grace changes recalculate its deadline. New settings affect later checks and alerts; hosted plan limits and target restrictions still apply.

Stop scheduled checks and alerts for one monitor, preserving its configuration and history; returns the paused monitor. Use this for a pause-only change, or update_monitor to combine a pause with configuration edits. For planned downtime that should still run checks, schedule maintenance through the app or REST API instead. Use resume_monitor to restart checking.

Restart checking a paused monitor and return its record with pending status. Use this for a resume-only change; use update_monitor for combined configuration edits. Its next check becomes due immediately and may create incidents or send alerts. Hosted active-monitor limits are checked again; existing history is retained.

Check a monitor immediately. Without region, records the result through the normal incident/alert workflow and returns {ok:true}; it can notify channels. With region from list_regions, returns {ok:true,region,result} for an unrecorded spot check without changing monitor history/status. Use get_monitor to read existing checks. Heartbeats have no outgoing spot check; hosted check credits apply and probe/network errors are returned.

Discover configured check locations and the maximum regions per monitor on this deployment. Use the returned ids with create_monitor/update_monitor.regions or run_check.region; do not guess ids. auto is always available. Explicit remote regions support HTTP/TCP only; hosted plans may restrict how many are used.

Permanently remove a monitor and its check history; returns an empty success response. Use pause_monitor instead when you may need it again. Requires workspace owner/admin access in addition to write scope. The monitor cannot be recovered through this tool; deletion does not create an incident-resolution update.

Find automatic and manually opened incidents visible in the selected project, with open incidents first. Returns incident summaries and UUIDs; use get_incident for the timeline, failed checks and investigation details. To change status or resolve an incident, use post_incident_update rather than update_incident.

Investigate one incident by its UUID from list_incidents. Returns its write-up, update timeline (including internal updates), related monitor/checks, assignee and notification deliveries. Use list_incidents to discover ids; use update_incident to edit the write-up and post_incident_update to change status. This is private team information, unlike list_public_status_page_incidents.

Open a manual incident in the selected project and return its record and UUID. Use this for an event not already represented by an automatic incident; use post_incident_update for an existing incident. A monitorId limits it to pages containing that monitor; otherwise it applies to project status pages. The optional body becomes the first investigating update, public unless visibility is internal. Creation assigns the caller; it does not send subscriber update emails.

Edit an incident's root cause or Markdown postmortem and return the updated record. Use post_incident_update instead to change status or notify subscribers. Omitted fields stay unchanged; null or blank text clears a write-up field. Write-ups remain private unless publishPostmortem is true; a published postmortem appears on public status pages after resolution, and false hides it again.

Append an incident timeline update and change its status; returns the new update record. Use this to communicate progress or resolve an incident, rather than update_incident (write-up only). Public updates appear on status pages and are emailed to subscribers; internal updates remain team-only and send no subscriber email. resolved closes the incident and must be public. Repeating a call creates another update and may send another email.

Assign an existing incident to the authenticated caller and return the incident record. Use this when taking ownership of an investigation; it does not resolve the incident or post a subscriber update. Use post_incident_update to communicate status, and list_incidents to discover the incident UUID.

Read status-page records and UUIDs in the selected project for management. Use these ids with update_status_page; use get_public_status_page with a slug to read the published component/outage summary instead. This private list does not include pages outside the selected project.

Publish a status page at /status/ and return its record and UUID. Use list_monitors to select components and an unused slug (a conflict returns an error); omitted monitorIds creates an empty page. The page and selected monitors' public status become readable without authentication. Use update_status_page to edit an existing page; branding options remain subject to the hosted plan.

Patch an existing status page by its UUID from list_status_pages; returns {ok:true}. Use this to edit presentation or selected components; use get_public_status_page to read the public result. Omitted fields stay unchanged; monitorIds replaces the complete component list and [] clears it. Changes affect the published page; an empty externalUrl removes its header link. Hosted branding restrictions still apply.

Discover alert-channel UUIDs and delivery types (email, webhook, Slack and others) in the selected project. Use the ids with create_monitor/update_monitor.channelIds to route alerts. Returns channel records, not a delivery test. Channel creation and secret configuration are managed through the app or REST API.

Read maintenance windows in the selected project, including schedule and monitor associations. Use this authenticated list for planning; list_public_status_page_maintenance only shows windows for a published page slug. This tool does not schedule downtime; use the app or REST API to create or remove windows.

Read the published health summary for a status-page slug without credentials. Returns overall/component status, 90-day uptime and active public incidents; use list_public_status_page_incidents for incident history or list_public_status_page_maintenance for downtime schedules. Only published information is returned; a missing page or one with MCP access disabled returns an error.

Read recent public incidents and their public updates for a published status-page slug, newest first. Use get_public_status_page for current health, or authenticated get_incident for internal investigation details. No credentials are needed; internal updates are excluded and a missing page or one with MCP access disabled returns an error.

Read scheduled and ongoing public maintenance windows for a published status-page slug without credentials. Use this to explain announced downtime; authenticated list_maintenance is the project-wide management view. Returns public window records; a missing page or one with MCP access disabled returns an error.