OpenDerisk

Why we included this project

OpenDeRisk is aimed at the on-call shift where you have to chase an incident's root cause through raw logs, traces, and code. It runs a small crew of cooperating agents, one for SRE triage, one that writes and runs analysis code, another for reporting, plus data and visualization helpers, and it works from alert data so the investigation stays guided instead of turning into a manual hunt. Everything is compiled into a visual evidence chain you can audit and share. The architecture is open and the agents are extensible, so the same plumbing can handle flame graph reviews or conversational analysis of fresh telemetry exports, not just the stock demo. Teams running their own monitoring and alerting stack get a self-hosted troubleshooting assistant that actually correlates signals rather than just collecting them.