danwo is the open-source testing platform for the agentic era. It clicks through your web and mobile app like a user, while emulated Stripe, Clerk, Resend and twenty more providers record every call, then breaks the run on purpose, once per fault you should survive.
POST /v1/checkout/sessionsGET checkout.stripe.com
POST /emailscheckout.session.completed
UPDATE creditsPOST checkout142msINSERT orders9msclick Pay262msGET checkout118mswebhook96mswebhook (dup)dupUPDATE credits×2expect Credits1.0sEmulated behind an egress proxy. Your app keeps calling the real hostnames.
Amazon S3
Auth0
Nylas
Nango
Unipile
Pusher
Novu
Tiptap
Composio
Gemini
Perplexity
Bedrock
AI GatewayYou write the happy path once. danwo records what it caused behind the app, then replays it under every fault your providers, network and database can throw.
Steps with locators, or plain English for the agent, plus an invariant: a rule that must hold whatever happened. danwo record can draft both.
An egress proxy answers real provider hostnames from emulators. Every call, webhook and statement is logged, and the agent's choices are cached for replay.
It replays the clean run once per planned fault: declines, 500s, timeouts, duplicate or forged webhooks, double-submits, Postgres and Redis faults, time travel.
The page said "Thank you". Did Stripe charge once? Did the email send? Did the ledger add up? That is where apps break, and where danwo looks.
Stateful stand-ins for the 22 providers your app pays for, with real hostnames, signed webhooks, sockets and the error shapes your code has to handle, for your frontend, API and workers together behind one proxy. No sandbox keys, no rate limits, no flakes.
Stripe
ResendAmazon S3
Google sign-in
Gmail
Calendar
Drive
Microsoft
Slack
GitHub
Twilio
Paystack
Flutterwave
BachsAuth0
ClerkNylasNangoUnipilePusherNovuTiptapComposio
OpenAI
AnthropicGeminiPerplexityBedrockThe faults the test's own steps could meet, planned from what they called, grouped and kept within a budget.
The agent's choices are cached as locators. Reruns are deterministic, fast and free.
Steps, screenshots, a provider waterfall, the effects diff and emulator state, secrets redacted.
The same tests, invariants and matrix on React Native and Expo apps for Android and iOS, in-app checkout browsers included.
checkout.stripe.comClaude Code, Cursor and Codex run, explain and write danwo tests through MCP, then fix what the matrix found.
event.id; the matrix now passes 18 of 18.Every pull request gets a check run and a comment: what broke, under which fault, with the flight record one click away.
An emulator is only worth what it matches. danwo compares every emulator with the real API by structure, from the provider's published description and its live answers, so a green run means what it says.
POST /v1/checkout/sessionsmatched livePOST /emailsmatched livePUT and GET an object on Tigrismatched livesign-in by emailed codematched liveGET /transaction/verifymatched liveTypeScript, a Playwright-shaped API, and an invariant that judges the backend rather than the pixels.
screen.getByRole with agent.act; both replay without a model.import { type StripeFixtures, test as base } from "@danwo/sdk"import { grantedCredits, paidCredits } from "../src/ledger.ts"const test = base.extend<StripeFixtures>()test("buys 500 credits through Stripe Checkout", async ({ app, screen, expect }) => { await app.open("/") await screen.getByRole("button", { name: "Buy 500 credits" }).click() await expect(screen.getByTestId("checkout-total")).toContainText("$5.00") /* Paying calls Stripe; readiness waits for the call and plays the webhook it causes. */ await screen.getByTestId("checkout-pay").click() await expect(screen.getByRole("heading", { name: "Thank you for your purchase" })).toBeVisible() await expect(screen.getByText(/^Credits: /)).toHaveText("Credits: 500")}).invariant(async ({ app, stripe, expect }) => { // Whichever way the purchase went, the shop grants exactly the credits Stripe was paid for. expect(await grantedCredits(app.baseUrl, "u1")).toBe(await paidCredits(stripe, "u1"))})UI-only AI testing tells you the button worked. danwo tells you what the button did.
danwoThe whole engine runs on your machines: runner, emulators, matrix, recorder and dashboard. The cloud adds history and a team around it.
Everything that runs a test, on your laptop or in your CI. Nothing leaves your machines.
danwo serve@danwo/playwrightThe same engine, with the history your team needs to trust it.
Not open yet. Everything above runs self-hosted today.