Drops into any failing CI pipeline, reads the logs and git history directly from the runner, and delivers a structured Root Cause Analysis to your team — so you know exactly what broke without ever opening the runner logs.
uses: ChengaDev/opsagent@v0.1.0
Everything you need to go from failure to root cause in seconds.
Runs as the last step of a failing workflow. No sidecar, no external service — reads logs and git history directly from the runner disk.
LangGraph orchestrates a structured 4-phase investigation: log triage → code correlation → workspace inspection → RCA synthesis.
Works with Anthropic (Claude), OpenAI (o4-mini), and Google (Gemini 2.5). Switch providers with a single line in your workflow.
Posts the RCA to Slack, Discord, Teams, PagerDuty, or any webhook. Also comments directly on the failing pull request.
A fast regex-based scanner runs before the LLM — extracts structured issues in milliseconds at zero token cost.
pytest, Jest, Docker, Helm, Terraform, Kubernetes — if it produces a log file, OpsAgent can analyse it.
From unit tests to infrastructure — OpsAgent speaks your stack.