Get started in 2 minutes
Install Claudeye and get instant control over every agent run.
What you can do
Browse sessions
View every Claude Code and Agents SDK session with full execution traces, tool calls, and subagent logs.
Set up evals
Grade sessions automatically using pre-built evals or your own custom logic. Get pass/fail results and 0-1 scores.
Alerting
Fire Slack webhooks, console warnings, or any async callback the moment sessions fail your quality checks.
Deploy your Reliability Agent
Run Claudeye as a sidecar to monitor your production agents. Catch failures at runtime and course correct before they reach users.
How it works
Claudeye reads JSONL session logs written by Claude Code (~/.claude/projects) and the Claude Agents SDK. It parses them locally. Nothing leaves your machine.
When you open a session, Claudeye runs any evals and enrichments you’ve registered and caches the results. Subsequent views are instant.
