Incident Library
Incident Library turns post-incident reports that organisations published openly into structured cases: symptoms, trigger, root cause, affected components, contributing factors, impact, timeline, detection, remediation and lessons. Every case links to the original report and an archived copy. Describe what you are seeing in a sentence, such as "pods stuck Pending after a node upgrade" or "TLS errors right after midnight", and Claude finds the real incidents that match, then reads the full case and its source. It can also answer "what usually causes this" by counting root causes across the matching cases, list cases by root cause, component, symptom, contributing factor or organisation, and turn plain words into the library's vocabulary. Five read-only tools: search_incidents, get_incident, list_incidents, get_topic and vocab. No account and no key; nothing is written. Cases are reviewed before they are published, in English with Korean text where reviewers wrote it.
Details
https://incidentlib-mcp.benchlet.com/mcp · streamable-http
