- collapse run log data blocks whose pretty-printed JSON exceeds a byte or line limit, with a per-block toggle and an expand/collapse-all button next to the Logs heading - align data blocks and toggles with the context/message column - add a "Copy logs" button that copies the filtered logs as plain text with every data block expanded - add GET /manifest.json with the app name, version and run log data limits (RUN_LOG_DATA_COLLAPSE_BYTES / RUN_LOG_DATA_COLLAPSE_LINES) - load the manifest in the client stale-while-revalidate: localStorage copy first, refetched per page load and on tab focus after a while - show the server version in the sidebar footer, with a reload hint when it differs from the client build Closes #3
5.6 KiB
liqa
liqa is a browser automation and QA service for web applications. It stores scenarios (ordered Playwright steps), snippets (reusable code run via context.runSnippet), environments, credentials and sessions, and runs scenarios in headless Chromium. Scenarios and snippets are versioned; each run records the scenario version it executed.
It is an npm workspace with two packages:
server/- NestJS API + scheduler + MCP server, SQLite via TypeORM; the schema is managed by migrations inserver/src/database/migrations/(see "Database migrations" indocs/development.md). Any entity change needs a generated migration -test/migrations.spec.tsfails otherwise. REST API under/api/v1, Swagger UI at/docs, MCP at/mcp, health at/health, public client configuration at/manifest.json.client/- React 19 + Vite SPA (i18n strings inclient/src/i18n/locales/en.json).
Docs
- README.md - overview
- docs/architecture.md - services and modules
- docs/development.md - local setup, running, testing, Docker
- docs/scenario.md - scenario execution, script
contextAPI, versioning, export/import, Playwright traces - docs/mcp.md - MCP tool reference
- CHANGELOG.md - release notes per version
- Dockerfile - single image: builds client and server, runtime serves both
- .env.example - all configuration variables (validated in
server/src/config/app.config.ts)
Default deployment (docker compose)
compose.yaml at the repo root runs a single liqa container, built from the root Dockerfile, at http://localhost:13000. The server serves everything: the UI at / (the built client, via CLIENT_DIR), the REST API at /api/v1, MCP at /mcp, Swagger at /docs. It reads .env if present (optional - every setting has a default) and mounts ./data (SQLite DB data/liqa.db + stored files).
Rebuild and restart: docker compose up -d --build. Check with docker compose ps, curl localhost:13000/health and docker compose logs liqa.
The live database is data/liqa.db. The container applies pending migrations before starting the server, so back it up before deploying a release with new migrations.
Branching
Never commit directly to master when implementing a feature, chore or bug fix: create a separate branch for the work first. Only version-bump release commits (see "Committing a new version") are made on master.
Code quality
Before committing, run and pass all of these (and report the results):
# server
cd server && npm run lint && npx tsc --noEmit -p tsconfig.json && npm test && npm run build
# client
cd client && npm run lint && npm run build && npm run test:storybook
- Lint must stay clean in both packages (no errors, no warnings).
test:storybookruns every story inclient/.storybook/stories/in headless Chromium (needsnpx playwright install --only-shell chromiumonce, and again after Playwright updates). Add or update a story when adding or changing aclient/src/ui/component.- Don't run prettier over whole existing files; several aren't prettier-clean and it produces unrelated diffs. Format only new files or the lines you changed.
- Server tests use an in-memory DB, a Playwright mock (
server/test/__mocks__/playwright.ts), and a temporary files directory per test app (seeserver/test/app.harness.ts). - Import Node.js built-in modules with the
node:prefix (node:fs,node:path,node:crypto, ...).
Committing a new version
When asked to commit:
-
Ask which version to bump - major, minor or patch - and propose a sensible default based on the changes: patch for fixes and small behavior/default tweaks, minor for new features, major for breaking changes (e.g. changed API or export/import formats that old clients or files can't use).
-
Run the code quality checks above.
-
Bump
versionin the rootpackage.jsonand in both places at the top ofpackage-lock.json(the root entry andpackages[""]).server/package.jsonandclient/package.jsonkeep their own versions; don't bump them. -
Add a
## <version> - <YYYY-MM-DD>entry at the top ofCHANGELOG.md("Changes since :", then### Added/### Changed/### Fixed). Mark breaking changes with Breaking:. -
Update the docs affected by the change (
docs/*.md,README.md,.env.example, MCP tool descriptions). -
Build the commit message with the
/git-commitskill if it is available, following its format and rules. Otherwise use a conventional commit message (type(scope): subject, then a bullet list body). Either way, put the version bump in the body as a bullet (e.g.- bump version to <version>), not in the subject line. This overrides the skill's rule of putting versions in the subject. Then make the commit with that message. -
Tag the version commit with an annotated tag
v<version>whose message is that version'sCHANGELOG.mdentry, verbatim (from its## <version>heading up to the next version's heading). Commits without a version bump get no tag.--cleanup=whitespaceis required: the default cleanup strips lines starting with#, which would drop the entry's headings:v=<version> awk -v v="$v" 'index($0, "## " v " ") == 1 {p = 1; print; next} p && /^## / {exit} p' CHANGELOG.md \ | git tag -a "v$v" --cleanup=whitespace -F -Don't push the tag; tell the user to push with
git push --follow-tags, which also pushes annotated tags. -
After committing, ask the user whether to rebuild and restart the docker compose containers (
docker compose up -d --build); don't do it without confirmation.