Raindrop – Local Agent Debugger

Notifications You must be signed in to change notification settings

Fork 4

Star 182

BranchesTags

Open more actions menu

Folders and files

NameName

Last commit message

Last commit date

Latest commit

History

21 Commits

21 Commits

app

app

bin

bin

docs/assets

docs/assets

drizzle

drizzle

examples

examples

scripts

scripts

skills

skills

src

src

.gitignore

.gitignore

LICENSE

LICENSE

README.md

README.md

bun.lock

bun.lock

bunfig.toml

bunfig.toml

drizzle.config.ts

drizzle.config.ts

eslint.config.mjs

eslint.config.mjs

install.sh

install.sh

latest.json

latest.json

package.json

package.json

tsconfig.eslint.json

tsconfig.eslint.json

tsconfig.json

tsconfig.json

Repository files navigation

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.

Install

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

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.

Instrument your agent

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.

What it does

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.

Compatible with everything

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

Configuration

Env var Purpose Default

RAINDROP_WORKSHOP_PORT HTTP + WS port 5899

RAINDROP_WORKSHOP_DB_PATH SQLite database file ~/.raindrop/raindrop_workshop.db

RAINDROP_LOCAL_DEBUGGER SDK-side: where to mirror traces unset

CLI

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

License

MIT.

About

Give your coding agent the power to write and run agent evals.

www.raindrop.ai/workshop/

Topics

tracing

raindrop

llm

Resources

Readme

License

MIT license

Uh oh!

There was an error while loading. Please reload this page.

Activity

Custom properties

Stars

182 stars

Watchers

0 watching

Forks

4 forks

Report repository

Releases 7

v0.1.6

Latest

May 14, 2026

+ 6 releases

Packages 0

Uh oh!

There was an error while loading. Please reload this page.

Contributors

Uh oh!

There was an error while loading. Please reload this page.

Languages

TypeScript 92.1%

Shell 4.4%

JavaScript 2.8%

Other 0.7%