
Agent K
Jul 2026When a SigNoz alert fires, Agent K investigates the failure through the SigNoz MCP server, forms root-cause hypotheses, and publishes only the claims it can back with a resolvable trace, log, metric or deployment link. Rollbacks run only if a code-level policy gate allows them, and the agent records its own tokens, cost and verdicts as spans, so its reasoning is auditable in the same dashboard as the app. Built for the Agents of SigNoz hackathon.













