Workshop by Raindrop
Give Claude CodeCodexDevinCursorOpenCodeClaude Code the power to write and run agent evals.
Free. Local. Open-source.
the local debugger your agent is missing.
macOS, Linux, and probably Windows.
01
Live streamed traces
Every token, every tool call, every decision - streaming into Workshop the moment it happens. No polling, no refreshing.
02
Integrates with your coding agent
Claude Code reads your traces, writes evals, and fixes what's broken.
03
Self-healing eval loop
Claude writes the eval, runs your agent, sees the failure, fixes the code, and re-runs - until every assertion passes.
On X
what builders are saying.
This is actually incredible. I've been building out a project that involved running Pi in a cloudflare container, it's been a nightmare locally to try and get decent observability on what's happening and to get a feedback loop going. This really came at the perfect time!
I've been using this all day today. Soooo much easier to iterate. Spans, token count, cost, timing, a sane view of each tools input-output.
finally somebody built this
loved the onboarding experience
nice, it works awesome for voice agents too!
Congrats on the launch! Have been building many versions of this from scratch for various projects, and now I can finally use a unified tool for it.
Very good. Immediately useful. Going to be using this as default for local agent work now Also looking at instrument Claude code sessions with it to improve skills etc
342 left
Limited Edition
get the exclusive launch drip.
Install Workshop and run the magic command. Free merch shipped worldwide while supplies last.