RELATED 01
PostHog MCP Server
OFFICIAL SERVERPostHog's hosted MCP endpoint for analytics queries, feature flags, experiments, errors and more
41kGitHub stars
40popularity score
Better Stack MCP Server
Hosted Better Stack server for incidents, on-call, uptime monitors, logs, metrics and error tracking
Official · betterstack.com Remote (hosted) · Streamable HTTP DevOps and monitoringData and analytics
Connect your client
Remote (hosted)
Add in one click
Opens the app, which asks before it adds the server.
Or add it by command or config
Claude Codeterminal
claude mcp add --transport http betterstack https://mcp.betterstack.comCursor~/.cursor/mcp.json
{
"mcpServers": {
"better-stack": {
"url": "https://mcp.betterstack.com/"
}
}
}Claude (desktop and web)Settings → Connectors → Add custom connector
https://mcp.betterstack.com/What it does
Better Stack MCP Server is the hosted endpoint Better Stack runs at https://mcp.betterstack.com so an assistant can work on a Better Stack account. It covers the company's Incidents side (incidents, on-call calendars, escalation policies, uptime monitors, heartbeats and status pages), its Telemetry side (log sources, ClickHouse SQL queries over logs, spans, metrics, exceptions and session replays, dashboards, charts, chart alerts and saved explorations), error tracking (applications, releases, error patterns, and links to Linear or Jira issues), team membership, and a search over Better Stack's own documentation.
The tool reference on its docs page names 120 tools. Many of them change things: creating and resolving incidents, creating monitors and heartbeats, editing status pages, building or deleting dashboards and charts, inviting or removing team members, and creating temporary credentials for direct ClickHouse access. All tools are exposed by default; an `X-MCP-Tools-Only` header (allowlist) or `X-MCP-Tools-Except` header (blocklist) narrows the set per client, which the docs suggest for read-only use.
Nothing runs on your machine. Clients that speak remote HTTP connect directly and sign in through the browser with OAuth, or send a Better Stack API token as a Bearer header. The endpoint's OAuth metadata names betterstack.com as the authorization server and lists two scopes, read and write (https://mcp.betterstack.com/.well-known/oauth-protected-resource). Better Stack also publishes a Claude Code plugin that points at the same endpoint and adds query and dashboard skills.
What it needs
A Better Stack account with the products you want to reach (Incidents, Telemetry, Error tracking). An MCP client that supports remote HTTP servers, signing in with OAuth in the browser or with a Better Stack API token in an Authorization header. Clients without remote HTTP support need the mcp-remote proxy, run with npx (Node.js). In Claude Code, run /mcp and sign in after adding the server. Claude Tag in Slack needs an admin to set it up in an Access bundle first.
Limits
Hosted only: there is no package or source to run yourself, and the docs give no self-hosted option. The docs list tools by readable labels ("List monitors"), while the tool filter headers take the tools' internal names (examples given: monitors, monitor, incidents, remove_dashboard, remove_chart); the docs do not publish the full list of internal names. Reopen incident works only within 24 hours of resolution. Removing a release does not stop it coming back: a release that still receives errors is detected again with its next event. Linking and creating issues from errors works with Linear and Jira through the team's existing integration.
Who it suits
Teams already on Better Stack who want an assistant to triage incidents, check who is on call, query logs and errors, or build dashboards without leaving the chat. Because it is official and hosted, there is nothing to install or update, but your data and queries go through Better Stack's endpoint. The default tool set can delete dashboards, charts and alerts and change team membership, so anyone who wants read-only access should set X-MCP-Tools-Only or X-MCP-Tools-Except. Not useful without a Better Stack account; for other monitoring vendors, pick that vendor's own server.
Tools 60 of 120
Search documentation- Searches Better Stack's docs and returns matching articles
Get on-call details- Shows who is on call for a calendar now or at a given time
List on-call calendars- Lists on-call calendars and their rotations
List on-call events- Lists upcoming shifts and schedule changes
List escalation policies- Lists escalation policies and their steps
List severities- Lists incident severity levels
Create incident- Opens a new incident
List incidents- Finds incidents by date, status, monitor and other filters
Get incident details- Returns one incident with its error message and screenshots
Get incident timeline- Returns an incident's full history
Create incident comment- Adds a comment to an incident
Acknowledge incident- Acknowledges an incident so escalation stops
Escalate incident- Escalates an incident to a user, team, schedule or policy
Resolve incident- Marks an incident resolved
Reopen incident- Reopens an incident resolved in the last 24 hours
List monitors- Lists uptime monitors with filters
Get monitor details- Returns a monitor's settings and status
Create monitor- Creates an uptime monitor for a site, host or service
Get monitor availability- Calculates uptime percentage and SLA summary
Get monitor response times- Returns a monitor's response time figures
List heartbeats- Lists heartbeat monitors and their status
Create heartbeat- Creates a heartbeat for a cron job or background task
Get heartbeat availability- Returns uptime for a heartbeat
List status pages- Lists your status pages
Update status page- Changes a status page's name, theme, layout or domain
Add status page resource- Puts a monitor, heartbeat or group on a status page
Remove status page resource- Takes a resource off a status page
Create status page report- Posts an incident or maintenance report on a status page
Create status page report update- Adds an update to an open status report
Execute query- Runs ClickHouse SQL over logs, spans, metrics, exceptions and replays
Render chart- Runs a query and draws the result as a line, bar or pie chart
Plan query windows- Estimates rows read and splits a long time range into windows
List sources- Lists log sources and their settings
Create source- Creates a log source
Get source fields- Lists the fields available to query in a source
List dashboards- Lists dashboards
Get dashboard details- Returns a dashboard's charts, sections and layout
Create dashboard- Creates a dashboard, optionally from a template
Export dashboard- Exports a dashboard's configuration as JSON
Import dashboard- Creates a dashboard from JSON
Remove dashboard- Deletes a dashboard and its charts
Add chart to dashboard- Adds a SQL-backed chart to a dashboard
Configure chart- Changes a chart's name, type, query or settings
Remove chart- Deletes a chart
Create chart alert- Adds a threshold, relative or anomaly alert to a chart
Remove chart alert- Deletes a chart alert
Pause or resume chart alert- Pauses or unpauses a chart alert
List explorations- Lists saved Explore queries
Get metrics schema- Lists available metrics with data point and series counts
Create metric expression- Extracts a metric or label from log fields
Create cloud connection- Issues temporary credentials for direct ClickHouse access
List applications- Lists error tracking applications
Create application- Creates an error tracking application
List releases- Lists an application's releases
List errors- Lists error patterns with counts, affected users and state
Get error details- Returns an error's type, message, call site and state
Update error state- Marks an error resolved, ignored or unresolved
Create issue for error- Opens a Linear or Jira issue for an error and links it
Invite team member- Invites someone to a team by email
Remove team member- Removes a team member or cancels an invitation