The local debugger your agent is missing. Watch your agent think locally, the moment it happens: every token, every tool call, every decision.

Give Claude Code the power to read your traces, write evals against your codebase, and fix what's broken.

One command. This is all you need to install and run Workshop. There is nothing to clone and nothing to build.

curl -fsSL https://raindrop.sh/install | bash

Using an AI coding agent? Tell it to run the command above. Do not clone or build from source just to try Workshop; that path is only for people developing Workshop itself.

Open your coding agent of choice in your repository and run:

/instrument-agent

This will instrument your agent with Raindrop tracing and open Workshop in your browser.

That's it. Traces stream into the UI the moment your agent runs.

• Live streamed traces. Every token, tool call, and span streams into Workshop as it happens. No polling, no refreshing.
• Coding-agent integration. Claude Code reads your traces, writes evals against your codebase, and fixes what's broken.
• Self-healing eval loop. Claude writes the eval, runs your agent, sees the failure, fixes the code, and re-runs until every assertion passes.
• Local replay. /setup-agent-replay scaffolds an HTTP endpoint that replays a production trace against your real agent code.

Workshop is the local debugger. Raindrop Cloud is the hosted product: production observability for your AI features at app.raindrop.ai. The same raindrop binary connects your project to it, with no local daemon involved.

Connect a project to the cloud:

raindrop cloud setup

This signs you in (opening a browser the first time; see raindrop login below), writes your org's RAINDROP_WRITE_KEY to ./.env, and installs the hosted MCP server plus the cloud skills (raindrop-setup, raindrop-investigate) into your AI coding agents. Then run /raindrop-setup inside your agent to instrument your app, and events stream to app.raindrop.ai.

Sign-in is handled separately and reused across projects:

raindrop login    # OAuth sign-in; caches credentials in ~/.raindrop
raindrop logout   # clear stored credentials

raindrop cloud setup calls login for you only when you are not already signed in, so day-to-day you just run cloud setup.

When you run raindrop setup interactively, it also offers to set up Raindrop Cloud as an optional last step. Decline to stay Workshop-only, or accept to sign in and connect the project. Non-interactive runs (CI, piped scripts) never prompt; use raindrop cloud setup or the --cloud one-liner for those.

To undo a cloud install, run raindrop cloud uninstall. It removes the hosted MCP server and the cloud skills from your agents and clears the cloud install registry, leaving your local Workshop install untouched. Add --wipe to also remove RAINDROP_WRITE_KEY from ./.env.

The install one-liner takes a --cloud flag that connects the project to Raindrop Cloud instead of starting the local daemon:

curl -fsSL https://raindrop.sh/install | bash -s -- --cloud

Without --cloud the installer runs raindrop setup and the local Workshop daemon (the default above). With --cloud it runs raindrop cloud setup and starts no daemon. Local Workshop and Raindrop Cloud coexist: they use distinct MCP server names (workshop vs raindrop) and separate install registries, so neither overwrites the other.

• Languages: TypeScript, Python, Go, Rust
• SDKs: Vercel AI SDK, OpenAI Agents SDK, Anthropic SDK, Claude Agent SDK, LangChain, LangGraph, CrewAI, Mastra, Pydantic AI, DSPy, Google ADK, Strands, Agno, Deep Agents
• Providers: AWS Bedrock, Azure OpenAI, Vertex AI
• Coding agents: Claude Code, Codex, Devin, Cursor, OpenCode

raindrop workshop          # start and open UI
raindrop workshop setup    # write .env, then start and open
raindrop workshop status   # check health
raindrop workshop reset    # delete local DB after confirmation
raindrop update            # update the binary

raindrop login             # sign in to Raindrop cloud (OAuth)
raindrop logout            # clear stored credentials
raindrop cloud setup       # connect this project to Raindrop cloud (no daemon)
raindrop cloud uninstall   # remove the cloud MCP + skills from your agents

You only need this if you're hacking on Workshop itself. To use Workshop, run the install command above. You do not need to clone or build from source.

git clone https://github.com/raindrop-ai/workshop.git
cd workshop
bun install
bun run dev

bun run dev starts the local Workshop daemon and Vite UI. Open http://localhost:5899 after it starts.

MIT.