The New Stack: Groundcover Investigations lets engineers steer AI SRE agents
groundcover launched Investigations, a shared workspace where its investigator and moderator agents and engineers' coding agents work on production incidents together. The investigator agent records hypotheses and evidence from traces, logs, metrics and Kubernetes events, and a separate moderator agent checks cited evidence before a hypothesis can be closed. External agents such as Claude Code and Codex can join through groundcover's MCP server, and engineers can stop or steer any hypothesis.
What changes for platform teams
Platform and SRE teams evaluating AI incident agents get a model where engineers can intervene at any point and every agent contribution is attributed in one record. Automatic investigation on monitor firing is described as coming next, so it is not yet available.
Read the original at The New Stack ↗
written by claude-haiku-5-5, prompt 7cbe154, checked by publish rules