Getting Started
From installation to your first passing case in about 10 minutes.
1. Install
- macOS: open the
.dmgand drag TestDog into Applications. - Windows: run the NSIS installer and follow the wizard.
The installer bundles a Node runtime and browser environment — end users do not need Node.js. Upgrades never overwrite your existing database; new columns are migrated automatically on startup.
For developers
To run from source: npm install → npm --prefix server install → npm run tauri dev. See "Development" in the repository README.
2. Configure the AI gateway
AI generation and self-healing need an LLM. TestDog talks to model services over the OpenAI-compatible protocol, so any compatible service works: DeepSeek official, OpenRouter, a company gateway/proxy, etc. Open Settings in the sidebar and fill in:
- Gateway URL / API key / model name (required, provided by your provider; DeepSeek's official URL is
https://api.deepseek.com) - Model has vision: sends image attachments and screenshots directly to the model (the model must support vision)
- Reasoning depth (reasoning models only; keep off for normal models)
See Settings for details.
3. Create a project and a case
- Open Projects and create a project with a start URL (the address under test; generation/recording/replay open it by default).
- Inside the project, create a test case with a title and description.
Does the system under test need login?
Record the login state once as a login config; generation and replay then run logged-in, without testing login in every case. See Test Data & Login.
4. Generate your first script
Three ways — pick any:
| Method | Best for | Docs |
|---|---|---|
| AI generation | You can only describe the flow in words, or interactions are complex | AI generation |
| Manual recording | You already know each step | Manual recording |
| Import .testcase | Shared case files / coding-agent output | .testcase files |
Recommended flow: AI-generate the first version → tweak steps manually → verify by replay → keep as a regression case.
5. Run the replay
Hit Run on a script version; Playwright replays deterministically and executes assertions. Results and failure diagnostics live under Run records. See Replay.