Getting Started

Cotracer prevents outages in distributed systems. It follows a code change through your production traces, logs, and code to the services that actually depend on it, and reports what will break — with the evidence behind every claim. Runs in CI or interactively on your machine, entirely inside your network, against your own model endpoint.

How it runs

Two pieces, both inside your network:

• Backend — a single container (Docker or Kubernetes). It runs the analysis, talks to your tracing, logs and GitHub, and calls your LLM. Nothing is sent to Cotracer: prompts, traces and code never leave your network.
• CLI (cotracer) — a thin client on CI runners and developer machines. It sends “analyze this PR” to the backend and prints the report. In CI its exit code fails the check on HIGH risk.

The only outbound calls to Cotracer are the registry pull (when you install or upgrade) and a daily license refresh — never in the analysis path.

What you need

• Your tracing and logs backends, reachable from the backend’s network
• Your code host, connected through a read-only app you create
• An LLM endpoint you control (cloud API or self-hosted)
• Docker Compose or Kubernetes for one container (1–2 GB RAM)

Which integrations your stack needs is settled in the onboarding call.

Path to your first analysis

1. Install — get access, pull the backend, start it.
2. Configure — point it at your tracing, logs and LLM.
3. Connect GitHub — create the GitHub App it reads code with.
4. Run it in CI — analyze every PR, or run the CLI locally.
5. Troubleshooting — when something doesn’t start.

Not in the pilot yet? Apply here.