---
title: Tools
description: Every tool Glowo's MCP server gives an agent, what it does, and the access it needs.
sidebar:
  label: Tools
---

{/* Rendered from packages/shared/src/mcp-tools.ts (MCP_TOOLS: names,
    descriptions, scopes) by components/McpTools.astro. The hosted server
    registers exactly these tools and gates each on these scopes, and
    packages/api/src/mcp/application/mcp-tool-authority.test.ts holds each
    tool's scopes equal to what its use case enforces. */}

These are the tools Glowo's MCP server at `https://api.glowo.dev/mcp` offers.
Your client shows the agent each tool's parameters, so this page says what each
one does and what it needs.

- **Needs** is the scopes a call must carry. A write or delete scope also
  covers the matching read.
- **Lowest level** is the narrowest access level on the
  [consent screen](/ai-agents/scopes) that grants them.
- An [API key](/ai-agents/connecting#use-an-api-key) needs the same
  capabilities, and must allow MCP.

Every connection sees every tool. Calling one it was not granted returns an
error naming the missing scope, and the agent is told to say so rather than
work around it.

## The tools

### Workspace

| Tool | What it does | Needs | Lowest level |
| --- | --- | --- | --- |
| `glowo_whoami` | Say which workspace this connection acts in, what it may do (an OAuth connection's scopes, or an API key's capabilities) and on whose behalf. Call it first. | Nothing | Read Only |

### Monitors

| Tool | What it does | Needs | Lowest level |
| --- | --- | --- | --- |
| `glowo_list_monitors` | List monitors in the workspace. Returns a slim payload (no dailyStats / heartbeats / flapState / config) — use glowo_get_monitor for full details. A non-null checkInterruption means the status shown is the last confirmed one and checks are interrupted or outage confirmation is paused. Supports limit/offset pagination. | `monitors:read` | Read Only |
| `glowo_get_monitor` | Get full details for a single monitor (config, current status, last heartbeat). A non-null checkInterruption means the status shown is the last confirmed one and checks are interrupted or outage confirmation is paused. | `monitors:read` | Read Only |
| `glowo_get_monitor_stats` | Workspace-wide monitor statistics: counts by status, uptime trends, recent incidents. | `monitors:read` | Read Only |
| `glowo_get_monitor_hourly_stats` | A monitor's uptime hour by hour over a period: up, down and degraded counts and average latency for each hour, across every region or only the ones named. | `monitors:read` | Read Only |
| `glowo_list_heartbeats` | A monitor's heartbeats (check results), newest first: status, latency, status code, region and message, over a period or for some regions if asked. Use it to spot flapping or recent failures, then call glowo_get_heartbeat_detail with a heartbeat's id for its timings, response and assertion results. | `monitors:read` | Read Only |
| `glowo_get_heartbeat_detail` | Full per-check detail for a single heartbeat: timings, response headers/body, assertion results, resolved IP. Use to triage a DOWN check. | `monitors:read` | Read Only |
| `glowo_create_monitor` | Create a new monitor. Tip: run glowo_probe_url first to validate the target before creating. Optional projectIds attaches the new monitor to one or more projects. | `monitors:write` | Operate |
| `glowo_update_monitor` | Update fields on an existing monitor (partial — only provided fields change). | `monitors:write` | Operate |
| `glowo_delete_monitor` | Permanently delete a monitor and all its heartbeats. Cannot be undone. | `monitors:delete` | Full Access |
| `glowo_pause_monitor` | Pause a monitor — suppress checks and alerts until resumed. For planned work, schedule glowo_create_maintenance_window instead: it ends on its own. | `monitors:write` | Operate |
| `glowo_resume_monitor` | Resume a paused monitor — checks and alerts re-enable immediately. | `monitors:write` | Operate |
| `glowo_run_check_now` | Dispatch an on-demand check across the monitor's regions. Returns `{correlationId}` immediately — the actual heartbeat appears via the normal check pipeline. Poll glowo_list_heartbeats (most recent first) shortly after to find the matching result; typical end-to-end latency is 2–10s depending on region count and TLS handshake time. | `monitors:write` | Operate |
| `glowo_acknowledge_monitor` | Acknowledge a monitor's alert while it is DOWN or DEGRADED: for the next 30 minutes no new alert for it is sent. Its recovery still is, and its checks keep running. Say in the message what is being done. | `monitors:write` | Operate |
| `glowo_probe_url` | Send one request to a URL from Glowo's servers (not from the check regions), without creating a monitor, and read the status code, headers, the start of the body, latency and how any assertions fared. Redirects are not followed. Use it to try a target and its assertions before glowo_create_monitor. | `monitors:write` | Operate |
| `glowo_acknowledge_dns_change` | Accept a DNS_DIFF monitor's current records as the intended ones: they become its baseline (with the records its recent checks saw) and the monitor reads UP again. | `monitors:write` | Operate |
| `glowo_list_monitor_acknowledgments` | Who acknowledged a monitor's most recent alerts, when, and the message they left, newest first. | `monitors:read` | Read Only |
| `glowo_get_monitor_report` | A monitor's SLA report for a period (from before to, ISO 8601): uptime percentage, incident count, average latency, time paused, maintenance time left out of uptime and daily stats. | `monitors:read` | Read Only |

### Projects

| Tool | What it does | Needs | Lowest level |
| --- | --- | --- | --- |
| `glowo_list_projects` | List all projects in the workspace. | `monitors:read` | Read Only |
| `glowo_create_project` | Create a new project to organize monitors. Slug must be URL-safe and unique within the workspace. | `monitors:write` | Operate |
| `glowo_add_project_monitors` | Attach existing monitors to a project. Idempotent — already-attached monitors are skipped. | `monitors:write` | Operate |
| `glowo_remove_project_monitor` | Detach a monitor from a project. The monitor itself is not deleted — only the membership is removed. | `monitors:write` | Operate |
| `glowo_delete_project` | Permanently delete a project and every sub-project under it, with their memberships. Monitors and status pages are not deleted — they only stop belonging to those projects. | `monitors:write` | Operate |
| `glowo_get_project` | Fetch one project with its monitor and status page counts. | `monitors:read` | Read Only |
| `glowo_update_project` | Rename a project, change its slug or description, or move it under another project (only the fields you send change). | `monitors:write` | Operate |

### Incidents

| Tool | What it does | Needs | Lowest level |
| --- | --- | --- | --- |
| `glowo_list_incidents` | List incidents in the workspace (open and recently resolved). | `incidents:read` | Read Only |
| `glowo_get_incident` | Get full details for an incident: status, affected components, timeline of updates. | `incidents:read` | Read Only |
| `glowo_create_incident` | Open an incident on a status page, to tell its readers about an issue yourself (monitors open their own incidents separately). Get the statusPageId from glowo_list_status_pages. Start with status=INVESTIGATING, then post updates as the situation evolves. | `incidents:write` | Operate |
| `glowo_add_incident_update` | Post a timeline update on an existing incident: new message + status. Subscribers are notified per the channel's settings. | `incidents:write` | Operate |
| `glowo_resolve_incident` | Resolve an incident, with an optional closing note (subscribers receive a resolution notification). With mode undo, takes back a reopening posted by mistake instead. | `incidents:write` | Operate |
| `glowo_reopen_incident` | Reopen a resolved incident, e.g. when the issue recurs. With mode undo, takes back a resolution posted by mistake instead. | `incidents:write` | Operate |
| `glowo_archive_incident` | Archive an incident — hides it from the status page timeline. Required step before glowo_delete_incident. Reversible until deletion. An incident the platform opened for a monitor (autoGenerated) must be resolved first, with glowo_resolve_incident: while open it is tracking a monitor that is still down. | `incidents:write` | Operate |
| `glowo_list_status_page_incidents` | List one status page's incidents, newest first. Archived incidents are left out; glowo_list_incidents searches the whole workspace. | `incidents:read` | Read Only |
| `glowo_update_incident` | Correct an incident's title, severity or affected components (only the fields you send change). On a published incident the change is public at once. | `incidents:write` | Operate |
| `glowo_edit_incident_update` | Reword an update already posted on an incident; find its updateId with glowo_get_incident. On a published incident the new wording is public at once. | `incidents:write` | Operate |
| `glowo_publish_incident` | Publish a DRAFT or SCHEDULED incident now: it appears on its status page and the page's subscribers are notified. | `incidents:write` | Operate |
| `glowo_schedule_incident` | Schedule a DRAFT incident to publish itself at scheduledFor (ISO 8601, in the future), typically announced maintenance. | `incidents:write` | Operate |
| `glowo_unarchive_incident` | Restore an archived incident to its status page's incident list. | `incidents:write` | Operate |
| `glowo_delete_incident` | Permanently delete an incident (and all its updates). The incident must be archived first via glowo_archive_incident. Use sparingly — for retracting an incident posted by mistake. Prefer glowo_resolve_incident when an incident actually happened. | `incidents:delete` | Full Access |

### Status pages

| Tool | What it does | Needs | Lowest level |
| --- | --- | --- | --- |
| `glowo_list_status_pages` | List status pages in the workspace. Required to discover the statusPageId you need for glowo_create_incident. | `status_pages:read` | Read Only |
| `glowo_get_status_page` | Fetch a single status page by id, including its components, theme, custom domain, and visibility settings. | `status_pages:read` | Read Only |
| `glowo_create_status_page` | Create a status page. Like one created in the dashboard, it is public at its address at once; pass visibility UNPUBLISHED to keep it private until glowo_publish_status_page. monitors lists the monitors it shows, by UUID. A customDomain serves the page only once its DNS is verified, from the dashboard. | `status_pages:write` | Full Access |
| `glowo_update_status_page` | Change a status page (partial: only the fields you send change). On a published page the change is public at once, including which monitors it shows. | `status_pages:write` | Full Access |
| `glowo_publish_status_page` | Make a status page public at its address, for anyone to read. | `status_pages:write` | Full Access |
| `glowo_unpublish_status_page` | Take a status page offline: it serves nothing until it is published again. Its incidents and subscribers are kept. | `status_pages:write` | Full Access |
| `glowo_delete_status_page` | Permanently delete a status page, with its incidents, postmortems and subscribers. Cannot be undone. | `status_pages:delete` | Full Access |

### Postmortems

| Tool | What it does | Needs | Lowest level |
| --- | --- | --- | --- |
| `glowo_list_postmortems` | List postmortems, newest first, filtered by status page, incident, status (DRAFT or PUBLISHED) or text. Archived ones are listed only with archived: true. Supports limit/offset; returns the total. | `postmortems:read` | Read Only |
| `glowo_get_postmortem` | Fetch one postmortem: its status page, incident, status and body or external link. | `postmortems:read` | Read Only |
| `glowo_create_postmortem` | Write a postmortem for a status page, as a DRAFT nobody else sees until glowo_publish_postmortem. kind INLINE carries a markdown body; kind EXTERNAL links to an http(s) page instead. An incidentId must be a resolved incident on the same page. | `postmortems:write` | Operate |
| `glowo_update_postmortem` | Edit a postmortem's title, slug, body or external link (only the fields you send change). On a published postmortem the edit is public at once. | `postmortems:write` | Operate |
| `glowo_publish_postmortem` | Publish a postmortem on its status page, for anyone to read. | `postmortems:write` | Operate |
| `glowo_unpublish_postmortem` | Take a published postmortem off its status page; it becomes a draft again. | `postmortems:write` | Operate |
| `glowo_archive_postmortem` | Archive a postmortem: it leaves the status page and default lists, and is kept (list it with archived: true). | `postmortems:write` | Operate |
| `glowo_link_postmortem` | Attach a postmortem to a resolved incident on the same status page. An incident carries one postmortem at most. | `postmortems:write` | Operate |
| `glowo_unlink_postmortem` | Detach a postmortem from its incident; the postmortem is kept. | `postmortems:write` | Operate |
| `glowo_delete_postmortem` | Permanently delete a postmortem. Cannot be undone; archiving keeps it instead. | `postmortems:delete` | Full Access |

### Maintenance

| Tool | What it does | Needs | Lowest level |
| --- | --- | --- | --- |
| `glowo_list_maintenance_windows` | List maintenance windows scheduled for a monitor (past, current, and future). | `monitors:read` | Read Only |
| `glowo_create_maintenance_window` | Schedule a maintenance window for a monitor. Alerts are suppressed during the window and the public status page shows a maintenance banner. | `maintenance:write` | Operate |
| `glowo_delete_maintenance_window` | Cancel a scheduled maintenance window. Alerts resume immediately if the window was active. | `maintenance:write` | Operate |
| `glowo_update_maintenance_window` | Move or edit a scheduled maintenance window: its title, description, start, end, timezone, recurrence or whether it counts against uptime (only the fields you send change). | `maintenance:write` | Operate |

### Notifications

| Tool | What it does | Needs | Lowest level |
| --- | --- | --- | --- |
| `glowo_list_notification_channels` | List all notification channels (Slack, email, webhooks, etc.) in the workspace. | `notifications:read` | Operate |
| `glowo_create_notification_channel` | Create a notification channel (email, SMS, Slack, Discord, Telegram, PagerDuty, webhook). `config.type` picks the kind; each kind's fields are in this tool's input schema. | `notifications:write` | Operate |
| `glowo_update_notification_channel` | Update a notification channel — rename, swap config, change events, retarget monitors/projects. Partial: only set the keys you want to change. | `notifications:write` | Operate |
| `glowo_enable_notification_channel` | Enable a notification channel so it starts firing on its configured events. | `notifications:write` | Operate |
| `glowo_disable_notification_channel` | Disable a notification channel without deleting it — useful for muting alerts during a known issue or maintenance. | `notifications:write` | Operate |
| `glowo_test_notification_channel` | Send a test notification through a channel to verify routing/credentials. The message is real: it reaches the channel's destination. | `notifications:write` | Operate |
| `glowo_delete_notification_channel` | Permanently delete a notification channel and detach it from any monitors that referenced it. | `notifications:delete` | Full Access |
| `glowo_get_notification_channel` | Fetch one notification channel: its type, events, linked monitors and settings, with secrets masked. | `notifications:read` | Operate |
| `glowo_list_channel_deliveries` | A channel's recent deliveries, newest first: what was sent, whether it arrived, attempts and the last error. Destinations and signed links are masked. | `notifications:read` | Operate |

## Choosing a level

**Read Only** answers what is up, what is down and why, which is most of what
people ask an agent. **Operate** adds the day-to-day on-call work: creating and
changing monitors, running checks, acknowledging alerts, opening, updating
and resolving incidents, writing and publishing postmortems, scheduling
maintenance and managing notification channels. It cannot set up or change a
status page, and cannot delete anything. **Full Access** adds both.
