Raindrop Workshop - Open source, free, local debugger for AI agents
by•
Raindrop Workshop is the first local debugger for agents.
It's free, local, and open source.
Your local agent traces stream, token-by-token, instantly. Another Agent like Claude Code can read them over MCP. Then Claude can write evals, replay traces, fix bugs... and do it all over again.
This is the Self-Healing Agent loop. And it’s only possible on Raindrop.
Check it out and star on Github here: https://github.com/raindrop-ai/w...
Replies
Raindrop
Hey PH! This is Alexis, Co-Founder of Raindrop.
Your agent fails at 1am, traces are in some SaaS dashboard, harness is on your machine, eval suite is in a third place, and Claude sees none of it.
We've been stuck in this loop. So we built our way out of it. Workshop is the first sane way to debug your agent locally.
It has two parts: a local UI and an MCP.
Every span streams live to a local browser UI and you can replay any agent run with edited prompts, models, and tools.
The MCP lets you create self-healing eval loops. Claude Code reads your traces, writes evals, and fixes what's broken.
It's free, open source, and works with all the agent SDKs you already use.
One command to install: curl -fsSL https://raindrop.sh/install | bash
Excited to hear what you think!
Showtime Gasless NFT Marketplace
Welcome to the self-improving agentic world. Very solid team, great launch. Congrats Alexis!
Raindrop
@alexmasmej It's a crazy world! Thanks so much Alex
This was so easy to integrate and view my agent traces. No more console logging!!
Raindrop
@ryandonofrio huge!! love to hear it
Really excited about this! If I'm a fan of Vercel's AI SDK local debugger what should I expect from this one? Raindrop integration I assume?
Raindrop
@igel amazing! the biggest thing is that it exposes a local MCP so Claude can write evals, replay traces, fix bugs... and do it all over again in a self-healing loop
Sidekick
@igel this also allows you save traces, re-run from production, is compatible with pretty much any SDK (including Claude Code!) etc.
@benhylak Really really appreciate
Mailwarm
What’s the easiest way to hook Raindrop into an existing agent project?
Raindrop
@thamibenjelloun great question, you can actually just run the command below and then call a skill from your coding agent and it will set it up!
1 curl -fsSL https://raindrop.sh/install | bash
2 /instrument-agent in your coding agent
Let me know how it goes!
Reducto Studio
This is really interesting, excited to play with it
Raindrop
@aditabrm excited to get your thoughts!
This is incredible! Using it to improve your prompting and execution is already a massive improvement, but your agent being able to use it itself just tightens the feedback loop. Huge!
Raindrop
@paolodamico amazing to hear :)
We've been building an agent and this is exactly what we've been looking for! so excited to give this a spin
Raindrop
@yitong_zhang so excited to get your feedback!
Finally, a proper debugger for the agent era. Bridging traces over to other agents via MCP is such a smart architectural move. I love the focus on local-first—it makes iterating on sensitive agent logic much more viable. Does Raindrop support custom visualizers for specific tool-call outputs yet?
Raindrop
@vanshvardhan_sorte exactly! what are you looking for from the custom visualizer? it's also open source so we're excited to see what people contribute :)
Shortwave
Agents have been such a black box, you're all doing an amazing job and helping me make sense of building in the new world. Congratulations and thank you for this!
Raindrop
@jamestamplin thanks so much James!! Glad to hear its been useful :)