Labsco
MCP SERVER · OFFICIAL PROJECT

Runframe

by runframe

Run incidents, on-call, and postmortems from your editor.

Observability, Monitoring & Incident Response
Summary
The whole incident loop without the browser tab.

It covers an incident end to end — acknowledge, page a responder, log findings to the timeline, escalate, resolve, write the postmortem — from wherever the work is already happening. Because each tool is bound to a scope, you can give an agent a key that reads incidents and updates the timeline but cannot page anyone at 3am.

What it is

Runframe is an incident lifecycle platform covering incident response, on-call, and status pages. This MCP server translates tool calls into Runframe API requests scoped by your API key, so an incident can be handled from an IDE or an agent instead of a browser tab. The server is stateless and stores nothing locally.

What you get
  • Incident handling: `runframe_list_incidents` and `runframe_get_incident` read them, `runframe_create_incident` opens one, and `runframe_update_incident` edits title, description, severity, or assignment.
  • Moving an incident along: `runframe_change_incident_status` sets new, investigating, fixing, monitoring, resolved, or closed, and `runframe_acknowledge_incident` auto-assigns and starts SLA tracking.
  • Getting a human involved: `runframe_page_someone` pages a responder over Slack or email, and `runframe_escalate_incident` moves to the next level of the policy.
  • `runframe_add_incident_event` adds a timeline entry while the investigation is still running.
  • On-call lookups: `runframe_get_current_oncall` answers who is on call right now, and `runframe_get_escalation_policy` returns the policy for a severity level.
  • Postmortems: `runframe_create_postmortem` and `runframe_get_postmortem`.
  • Service and people lookups: `runframe_list_services`, `runframe_get_service`, `runframe_list_teams`, and `runframe_find_user`.
Requirements

A recent Node.js and a Runframe API key from your account settings, passed as `RUNFRAME_API_KEY`. The default stdio transport runs via `npx -y @runframe/mcp-server` with no network exposure; the streamable HTTP transport for containers and remote hosts additionally takes a bearer token. Every tool declares the API-key scope it needs, so a key can be narrowed to reads.

Setup effort

One command plus a key — claude mcp add runframe -e RUNFRAME_API_KEY=rf_your_key_here -- npx -y @runframe/mcp-server, then supply credentials