About RapidWorkers
RapidWorkers lets software and AI agents hire people for paid data collection, real-device testing, and human-judgment tasks. Its SDKs and MCP server support quoting, posting jobs, and reviewing and approving or rejecting submissions.
Key features
- 01Photo, audio, video, and text collection
- 02Real-device app and site testing
- 03Surveys and usability feedback
- 04Preview job costs before posting
- 05Review, approve, or reject submissions
Use cases
- 01Collect data to a task brief
- 02Test an app or website on real devices
- 03Gather native-speaker or usability feedback
- 04Review worker submissions and payments
MCP support
The MCP server exposes RapidWorkers task tools through a hosted endpoint and local stdio relays. MCP clients can use OAuth with the hosted server or configure a local server with an API key.
Configuration
Remote servers need no local install. Pick your client, then copy the config or install in one click.
https://www.rapidworkers.io/api/mcp/claude mcp add --transport http io-rapidworkers-rapidworkers https://www.rapidworkers.io/api/mcp/{"mcpServers":{"io-rapidworkers-rapidworkers":{"type":"http","url":"https://www.rapidworkers.io/api/mcp/"}}}Open in Cursor{"servers":{"io-rapidworkers-rapidworkers":{"type":"http","url":"https://www.rapidworkers.io/api/mcp/"}}}Open in VS Code{"mcpServers":{"io-rapidworkers-rapidworkers":{"type":"http","url":"https://www.rapidworkers.io/api/mcp/"}}}{"mcpServers":{"io-rapidworkers-rapidworkers":{"serverUrl":"https://www.rapidworkers.io/api/mcp/"}}}Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.
Tools
Details
From the project READMERapidWorkers SDKs and MCP server
RapidWorkers lets software and AI agents hire real people for small paid tasks done on their own devices:
- AI data collection: photos, audio, video and text gathered to your brief.
- Real-device testing: people install your app or open your site and report what broke, with screenshots.
- Human judgment: native-speaker checks, surveys, usability and first-impression feedback.
You get a cost quote, confirm it, post the task, then review each submission and approve (pay) or reject it. This repository holds the official clients. Every one of them talks to the hosted MCP server at https://www.rapidworkers.io/api/mcp/, so they all share the same tools, validation and safeguards.
Packages
| Language | SDK | MCP server for local clients |
|---|---|---|
| Python | pip install rapidworkers | uvx rapidworkers-mcp |
| JavaScript / TypeScript | npm install rapidworkers | npx -y rapidworkers-mcp |
| Rust | cargo add rapidworkers | cargo install rapidworkers-mcp |
| Go | go get github.com/RapidWorkers-io/rapidworkers/go | go install github.com/RapidWorkers-io/rapidworkers/go/cmd/rapidworkers-mcp@latest |
| Docker | docker run -i --rm -e RAPIDWORKERS_API_KEY ghcr.io/rapidworkers-io/rapidworkers-mcp | |
| Claude Desktop | One-click extension: download rapidworkers-mcp-<version>.mcpb from Releases and open it |
All of them need an API key from https://www.rapidworkers.io/dashboard/api-management/, passed as RAPIDWORKERS_API_KEY.
Connect an AI agent
The easiest way is the hosted server, with OAuth sign-in and nothing to install:
claude mcp add --transport http rapidworkers https://www.rapidworkers.io/api/mcp/
For clients that launch local servers (Claude Desktop, Cursor, Windsurf, Cline, Zed...):
{
"mcpServers": {
"rapidworkers": {
"command": "npx",
"args": ["-y", "rapidworkers-mcp"],
"env": { "RAPIDWORKERS_API_KEY": "your-api-key" }
}
}
}
Use it from code
from rapidworkers import RapidWorkers
rw = RapidWorkers()
quote = rw.preview_job_cost(kind="ai_task", pay_rate=0.25, quantity=100)
print(quote["total"]) # confirm the cost before creating anything
import { RapidWorkers } from 'rapidworkers';
const rw = new RapidWorkers();
const quote = await rw.previewJobCost({ kind: 'ai_task', pay_rate: 0.25, quantity: 100 });
See each package's README for full examples.
Rules
At least $0.05 per worker ($0.10 for AI tasks, $0.50 per survey response), at least 10 workers per job, and a $20 minimum deposit. Creating a job needs a quote_id from preview_job_cost and confirm: true. Approving pays the worker and cannot be undone; pending submissions are approved automatically after 2 days.
How this repository works
spec/tools.jsonis the exact tool list of the hosted server.scripts/codegen.pygenerates each SDK's typed methods from it; CI fails if a generated file is stale.- The MCP packages are thin stdio relays: they forward every request unchanged to the hosted server, so they never fall behind it.
server.jsonis the MCP Registry entry.- Releases are published from GitHub Actions; see RELEASING.md.
License
MIT
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.