StateWeave documentation | StateWeave Docs

Get started

StateWeave documentation

Test your AI agent on working copies of the apps it uses. You fill a sandbox with the data your agent needs, write tasks with checks through the apps' own APIs, and run your agent on fresh copies as often as you like. Your agent runs where it always runs; StateWeave serves the apps and records every call.

Quickstart

A blank sandbox, a little data, one task, and your agent run against it.

Tasks

What your agent is asked, and checks that judge it through the apps' public APIs.

Run your agent

Fresh copies of the apps for every attempt, from any language or harness.

From your coding agent

Claude Code, Codex or Cursor fills the sandbox, writes tasks, runs your agent and fixes the code.

Reference

Every

sw

command, the Python SDK and the REST API.

How it works

1

Make a sandbox

In your workspace, pick the apps. Each starts blank: the account, its owner and the owner's token.

2

Fill it

Write the data through the apps' own APIs, from a script or your coding agent, or describe it in a prompt. Then switch it to Test: that keeps a version.

3

Write tasks

What your agent is asked, and how to tell it did it right, checked through the same APIs.

4

Run your agent

Each attempt gets fresh copies of the version. Judge it by the task's checks, fix your agent, run again.

Uploading a task to your workspace is optional. An uploaded task is a test: every run of it is graded and kept, and you can compare versions of your agent on it. See Upload a task .

Your agent never runs on StateWeave. It stays on your laptop, in CI or wherever you deploy it, and reaches the apps over HTTPS, so you test the real thing: your prompts, your tools, your model and your harness.

Apps

Each app is a twin of the real service: the same endpoints, request and response shapes, errors and limits, so the vendors' own client libraries work against it with only the address changed.

App API your agent calls MCP server
Slack Web API Yes
Gmail Gmail API v1 Yes
Google Calendar Calendar API v3 No
Google Sheets Sheets API v4, with Drive API v3 for files No
Google Drive Drive API v3 No
Salesforce REST API v68.0 Yes
Shopify Admin GraphQL API 2026-10 Yes
Zendesk Support API v2 No
WhatsApp Cloud API and Business Management API (Graph v26.0) Yes
Microsoft Teams Microsoft Graph v1.0, and the Bot Connector API v3 for a bot Yes
Stripe API v1 Yes
HubSpot CRM APIs: records, search, associations, properties, pipelines, lists, owners and custom object schemas Yes
GitHub REST API (versions 2022-11-28 and 2026-03-10) and GraphQL API: repositories, contents, branches, commits, issues, pull requests, reviews, releases, webhooks, teams and search Yes

See Apps and credentials for each app's addresses and client setup, and MCP servers for agents that use tools over MCP.

What you install

One package, stateweave-sdk , with no dependencies beyond Python's standard library:

sw , a command line: make and fill sandboxes, check and upload tasks, run your agent with sw run . See sw command line .

The stateweave Python module , the same from inside your own harness. See Python SDK .

Two skills for your coding agent (Claude Code, Codex, Cursor), added with sw setup : one connects your agent, the other makes sandboxes, uploads tasks and runs them. See From your coding agent .

Everything is also a REST API, for any other language. See REST API .

Get access

Create an account and name your workspace. It opens right away at its own address, such as https://acme.app.stateweave.net , where you make sandboxes, keep tasks and watch runs. Questions: support@stateweave.net .

Quickstart A blank sandbox, a little data, one task, and your agent run against it.

Tasks What your agent is asked, and checks that judge it through the apps' public APIs.

Run your agent Fresh copies of the apps for every attempt, from any language or harness.

From your coding agent

REST API

Upload a task

Apps and credentials

MCP servers

sw command line

Python SDK

Create an account