- feat: add `.agents/skills/playwright-cli/` with complete Skill markdown, agent interface, and 9 reference files covering session management, spec-driven testing, video recording, tracing, storage, request mocking, element attributes, test debugging, and running custom code - feat: add `.playwright/cli.config.json` to configure the Playwright CLI - feat: add `web/package.json` `playwright:cli` script that resolves `playwright cli` from the repo root - build: remove `@playwright/mcp` dependency from `web/package.json` - build: set `XDG_CACHE_HOME` in `mise.toml` - build: delete `.codex/config.toml` (MCP server config) - docs: update `README.md`, `docs/web/roadmap.md`, `docs/web/decisions.md`, and `web/README.md` to reflect the new Playwright CLI approach
MyGO Web
MyGO's browser client is a pure client-side rendered application built with React, TypeScript, and Vite.
Development
Install the repository-pinned Go and Node.js versions from the repository root:
mise install
Start the Go API from the repository root:
mise exec -- go run . serve
Create a development account through the existing public API if needed:
curl --request POST http://127.0.0.1:10086/api/v1/auth/register \
--header 'Content-Type: application/json' \
--data '{"username":"web-user","email":"web@example.com","password":"password123"}'
Then start the browser client from web/:
mise exec -- npm ci
mise exec -- npm run dev
Vite proxies /api to http://127.0.0.1:10086. Production deployments must serve the static build and /api/v1 from the same origin, normally through a reverse proxy.
Browser Testing and Debugging
The Web project uses the Playwright @playwright/test for both repeatable E2E tests and interactive browser debugging. The npm cache, browser binaries, CLI daemon state, screenshots, traces, and videos stay under the repository-level ignored .cache/ and .artifacts/ directories.
Install the Debian browser dependencies once per development container:
mise exec -- npm run playwright:install:deps
Install the matching bundled Chromium revision:
mise exec -- npm run playwright:install
E2E tests and interactive debugging intentionally share the same stable Playwright core and browser revision.
Currently Playwright's version is locked by the only @playwright/test dependency. Standalone @playwright/cli or @playwright/mcp won't be added unless additional requirement.
Run the headless browser tests with Chromium sandboxing enabled:
mise exec -- npm run test:e2e
Run interactive commands through the repository wrapper:
mise exec -- npm run playwright:cli -- open http://127.0.0.1:5173/login
mise exec -- npm run playwright:cli -- snapshot
mise exec -- npm run playwright:cli -- console
mise exec -- npm run playwright:cli -- requests
mise exec -- npm run playwright:cli -- screenshot --filename=.artifacts/playwright-cli/login.png
mise exec -- npm run playwright:cli -- close
The wrapper changes to the repository root before invoking the stable playwright cli bundled with the locked test runner. This loads .playwright/cli.config.json, which selects bundled Chromium in headless, isolated, sandboxed mode and limits generated output under .artifacts/playwright-cli/ to 100 MB.
Do not invoke playwright-cli, a global Playwright installation, or a floating npx package.
If the upstream CLI entry point changes, update only the playwright:cli npm script. Agents should follow the repository-scoped $playwright-cli Skill in .agents/skills/playwright-cli/ for snapshots, sessions, network diagnosis, test debugging, and artifact handling.
Checks
mise exec -- npm run check
The production build is emitted to dist/ as static assets. Browser E2E tests are run separately because they require the browser setup above. See docs/web/roadmap.md at the repository root for architecture boundaries and planned dependencies.