yellowjacket/.pi/skills/yellowjacket-dev/SKILL.md

name, description

name	description
yellowjacket-dev	Operating YellowJacket's development harness — which of the four test tiers to use for a given change, how to run the app headless and drive it with playwright-cli, seed and sandbox lifecycle, the three build-tag passes, and the failure modes that waste a cycle if you meet them cold. Use whenever building, running, testing or debugging this repo.

Working on YellowJacket

CLAUDE.md says what this system is. This skill says what to run. .planning/NOTES.md records what we measured and when. Keep them in those three tenses: if something here is wrong, fix it here and add the discovery to NOTES.md — do not add a corrective paragraph to CLAUDE.md.

Every command below is a make target on purpose. The Makefile is the source of truth for how to invoke something; this file only decides which and in what order. make skill-check fails if a target named here has disappeared.

Read this part before you fail

Fifteen things cost a cycle each the first time. They are here, not in a reference, because you need them before the failure, not after.

• Call a binding through the bridge. window.go does not exist under Wails v3 — the bindings are bundled modules, not a global — so use window.__yjEvents.call(path, args, ms) (browser) or callBinding (specs). Both post to the runtime's own endpoint by method name, so they work on any page, including one with no init script.

  A bad call now rejects, and says why: a wrong type comes back as a TypeError naming the argument, a wrong count as expects 4 arguments, got 3, an unknown method as a ReferenceError. Under v2 the backend logged error parsing arguments and never fired the callback, so .dev/app.log was the only place the reason appeared and the timeout was the only thing that made the mistake visible. The timeout is still there, but now it means a genuinely hung request.

• Nothing is clickable on a fresh YJ_HOME. <first-run-wizard> intercepts all pointer events until a library exists, and the click fails with a Playwright interception error that reads like a selector bug. Use a seed unless you are testing the wizard, in which case make dev-headless-fresh.

• Never pkill -f. The pattern matches the invoking shell's own command line, killing it and silently dropping the rest of your compound command. make dev-stop kills by saved PID.

• Seeds are produced by running the app, never by hand-writing a config.toml and DB rows — a hand-built YJ_HOME is a second description of a valid one and will drift. make sandbox-seed drives the real AddLibrary binding and waits for the real scan.

• …and a seed freezes every default it has already persisted. Changing a default in backend/config (or backend/tracklist) is invisible against an existing seed, whose config.toml holds the old value — while CI builds its seed by running the app and therefore tests the new one. Re-seed before believing either.

• A wa-dialog is awkward to locate, in three ways. The host is display: contents, so the element carrying your testid always reports hidden; the visible thing is the native <dialog> in its shadow root. The slotted content is in the host's shadow root, not in that dialog's subtree, so toContainText on the dialog sees only its chrome. And it has an accessible name only because utils/name-dialog.ts gives it one — Web Awesome does not wire label to aria-labelledby — so a new dialog that forgets to call the helper from updated() is invisible to getByRole('dialog', {name}).

• A name is computed on the element carrying the role, and Web Awesome puts the role in its own shadow root. aria-label on a <wa-slider> or a <wa-dialog> host never reaches the tree. Use the component's own label (plus styles/wa-slider-label.css.ts, since a slider's is visible) or utils/name-dialog.ts. And in the light DOM, a <label> that is a sibling of its control with no for names nothing — that was 24 of the 93 controls on Settings. getFullAXTree is how you check, and "0 unnamed" is not the whole answer: a placeholder is an accname fallback, so a box labelled only by one reports clean.

• The a11y snapshot cannot check an accessible name on a dialog. playwright-cli snapshot prints - dialog [ref=…] with no name whether the dialog is named by aria-labelledby, by aria-label, or not at all — checked all three ways against the running app. Use getByRole('dialog', {name}) in a spec, or CDP (Accessibility.getFullAXTree) for the browser's own computation, which also reports where the name came from. A snapshot read as a probe here reports failure on a working build.

• Playwright's WebKit does not run on Arch (Ubuntu-only libs). --browser=webkit is CI-only; local work is Chromium. CI runs it with if: !cancelled() so a chromium failure does not silently skip it, which it did for two sessions.

• CI's e2e job is green on both engines (88 specs each) since the container got an audio device that keeps time. If playback specs start failing there again, check the The sink plays at real time step first: ALSA's null plugin consumes 3000 ms of audio in 2.96 ms, so every track finishes instantly and the clock never moves — which reads as an app bug and cost two sessions of that suspicion.

• make e2e needs SEED=default. Its specs assert on fixture content — unicode tracks, the fixture artists, a known playable file. Run against the bulk seed a measurement session left behind and a third of them fail (13 of 36, when it was measured), in a list that reads exactly like a regression in whatever you are holding. make dev-headless SEED=default first.

• The catalog is stubbed out locally now, like CI. dev-headless.sh defaults YJ_CORE_INDEX_URL to a dead address because it was the only launcher that did not — seed-sandbox.sh and ci.yml always have. Without it the app downloads the real ~1M-row Explore catalog into the run's YJ_HOME, and specs that stage their own catalog rows then search a million real ones and fail locally only, which reads as a regression and is an environment. Pass YJ_CORE_INDEX_URL=<real url> when you want the real catalog to explore by hand.

• …and the suite spends state it cannot always give back. view-lifecycle.spec.ts skips an autotag album on every run, out of the eleven the seed has, and does not put it back — so around the eleventh consecutive run against one app it starts failing on an empty queue. Restart between runs (make dev-stop && make dev-headless SEED=default) when a spec starts failing that you have not touched, and before believing a failure at all. Backend state outlives the page: shuffle used to be left on the same way, which failed playback.spec on the next run — that one is fixed, the autotag one is inherent.

• A frontend edit is not live until you restart the app. Vite updates the module, but an already-registered custom element class cannot be re-registered, so a running page keeps the old one and your change reads as having done nothing — including across a browser reload. make dev-stop && make dev-headless SEED=…, then re-check. The nastier version: a build error leaves the dev server serving the last good bundle, so the page still works and still shows the old behaviour. make dev-headless prints the esbuild error; a reload does not. One way to cause one is a stray backtick inside a comment in a css tagged template literal, which ends the literal. That one is a check now — make css-check (instant, a pre-commit hook and a CI step) names the file, the line and the cause, because what you otherwise get is Property 'scroll' does not exist on type 'CSSResult' pointing at a line of prose, or every test in the suite failing to import. It went in after the trap cost a fourth session in which its own warning had been read twice.

• A failing CI job's log is reachable even when gitea_ci job_logs says it is not. That endpoint 404s on this Gitea build. The REST API answers, with the GITEA_TOKEN already in the environment: /api/v1/repos/yonlu/yellowjacket/actions/runs/<run>/jobs for per-step status (this is how "the WebKit step was skipped" was found) and /api/v1/repos/yonlu/yellowjacket/actions/jobs/<id>/logs for the whole log. Two sessions reasoned about the e2e failure from the commit list because the first tool's 404 read as "out of reach".

• npx tsc --noEmit is part of the gate, and nothing else runs it. CI does (.gitea/workflows/ci.yml), and it typechecks frontend/test/ — which make lint, make test, make ui-test and make e2e do not. A tree can be green on all four and red in CI.

Which tier

Four tiers. Start at the cheapest one that can see your change, and only climb when it cannot.

You changed	Run	Cost
A Lit component, a store, the shortcut service	make ui-test	~2 s, no app
…and it renders differently	make ui-visual	+ 6 baselines, opt-in
Any Go code	make test	3 passes, ~2 min
A service that emits events	make test — assert on the payload, see backend/queue/emit_test.go	in-process, no app
A bound method or a bound struct field	make bindings then make ui-test	~1.5 s + 2 s
A user-visible flow across frontend and backend	make e2e (needs the app up)	~1 min
Something you cannot predict — exploring	make dev-headless SEED=default + playwright-cli	interactive
Something whose answer is a number, not a pass	make perf against a bulk-seeded app	~1 min + setup
A .sql or .templ file	make generate, then the checklist in references/schema-change.md
Anything that has to survive on a phone	make android-smoke against a booted emulator	~1 min + setup

Two targets are once-per-clone prerequisites that are not dependencies of the targets needing them, so on a fresh checkout each fails with a missing-browser error that reads like a broken test: make ui-setup before make ui-test, and make e2e-setup before make e2e. (make testdata is a dependency of make test and make sandbox-seed; run it by hand only when invoking go test directly, since anything using internal/testfixtures skips rather than fails without it — a green run without the library means less than it looks.)

Two rules about climbing:

• A component test passing is not the app rendering. If you touched anything in frontend/src, verify it in the real app too — start it headless, screenshot --filename=/tmp/shot.png, and read the PNG. Two of this repo's worst regressions were only ever visible there: a header badge contradicting the settings page, and a virtualized row whose columns no longer lined up with its own header. Neither failed anything.
• A list that renders is not a list that repaints. lit-virtualizer re-renders its rows when one of its own properties changes, not when the parent does — so selection highlighting, the playing-track row and anything else driven by host state need an explicit virtualizer.requestUpdate(). Click a row and look, every time you touch one of these lists; the controller will hold the right state either way. Check el.viewActive first: dispatching a raw navigate event does not always activate a view, and an inactive one does not render at all — which looks exactly like this bug (the controller holds the selection, no row highlights) and is Phase 1 working as designed. Navigate by clicking the sidebar.
• Do not write an e2e spec first. Drive the flow by hand, then promote it with /e2e. Specs written blind assert on selectors that do not exist.
• Not every view has a nav item. Since #25 the destinations are configurable, Autotag is hidden by default and Downloads is absent until a download client exists — so getByTestId('nav-<view>') waits 30 s for a locator that will never resolve. navigateTo(page, view) (e2e/support/fixtures.ts) dispatches the app's own navigate event. Click the nav item when the nav is what the spec is about.

Before a commit, the gate is make lint, make test, make ui-test, make bindings-check, make css-check and — from frontend/ — npx tsc --noEmit. The first four are lefthook hooks, so skipping them locally only defers the failure; the typecheck is a hook too but only CI runs it over the test tree, which is where it has actually broken.

The message is gated too: make commit-check (a commit-msg hook, and a CI step over every commit in a push) rejects a subject that is not type(scope): subject, is over 72 chars, or ends with a period. A --no-verify commit skips it locally and meets it in CI.

Two things about the e2e tier that are not obvious until they bite. The 88 specs share one backend process in file order, so a spec that leaves the app somewhere passes alone and fails the suite — leave the UI as you found it, and wait for it rather than trusting the click to have finished. The queue panel's width is animated and the transport slides with it, so a click issued while it closes lands on whichever button moved under the pointer. And anything asserting on the queue panel's rows must open it first: a closed panel renders no list at all.

Measuring, when a pass is not the answer

Performance claims need a before and an after on the same machine against the same library, or they are anecdotes. The fixture library is a few dozen tracks and cannot show any of it.

make bulkdata                  # ~11 s, 466 MB into a gitignored .dev/
make sandbox-seed-bulk         # minutes: it is a real scan of 50 000 files
make dev-headless SEED=bulk
make perf LABEL=before         # ... make the change ...
make perf LABEL=after
make perf-compare BEFORE=before AFTER=after

Fourteen numbers: startup (and the count of cross-origin requests, which is whether the app works offline), the bundle's shape and each view's first open, keystroke-to-paint in the search box, what a naturally finished track provokes, what one favourite toggle costs, what sitting idle on Settings costs, what scrolling a long list costs (image bytes and the tier they were requested at, plus frame cost through the artist grid), what a long Explore session retains (heap sampled after each of twenty-four searches, plus every registered cache's size), what opening a 2 000-track playlist costs (elements retained, eager cover requests, heap, and what one update pass costs and rebinds), what the selection costs (ordering the selected keys with one row selected at either end of 50 000 and with all of them, and what "Select all → Edit tags" blocks for), what an update pass of the player bar costs (querySelectors, layout reads, style writes and the read-after-write interleaves inside updated(), measured with a clean DOM and a dirty one, plus six seconds of real playback), how many document pointer listeners are installed at rest (via CDP, so nothing else in the run is perturbed), what "play these" costs for an artist, twenty albums and five genres, and heap after a scripted browse. It wraps every bound Go method, so "did that refetch the library" is a fact rather than an inference.

window.__yjCacheStats() reports every registered cache's entries, retained chars and cap in one eval — which is how you check a bound is still holding without rebuilding the reproduction that justified it.

Adding a number is usually the first half of an item's work: most findings are not among the seven, and the fix cannot be believed without one. Two rules for adding one.

Stage what the seed does not have, idempotently and by name. The bulk seed has one empty playlist, against which "toggling a heart refetches every playlist" costs nothing and cannot be reproduced; the favourite measurement builds ten 500-track playlists first. Staging by name means a before and an after see the same shape — and dev-headless restores the seed tarball on every launch, so it is rebuilt each run anyway.

Measure both halves of a trade. Route splitting reports bytes before first paint and the slowest first open of a view, because a split that halves startup by making every page visibly slower has not helped anyone.

Measure the state the cost depends on, not just the operation. A forced layout costs 3 µs against a clean layout and 0.1 ms against a dirty one, so a component measured only in its steady state reports that the finding about it is imaginary. If the work is conditional, stage both conditions and put both rows in the table — they explain each other, and one of them is the number the fix has to move.

Fourteen traps, each of which produced a wrong number first:

• A label is a filename, and audit IDs are case-insensitive as filenames. .dev/perf/before-m6.json is the capital M6 (the 3 s ticker) from an earlier pass; measuring lowercase m6 under that name silently overwrites a baseline three passes of numbers depend on. Name a label after the change, not the finding.

• The first run after a rebuild is not a measurement — and the second is not reliably a good one either. A run taken immediately after make dev-headless often reports first contentful paint at 96–112 ms against 28–32 ms on the next run of the same build (a cold Vite module graph). But the ordering does not hold: one pass saw 100 then 96, and another 28 then 76. FCP moves ±50 ms for reasons this harness does not control, so take two, and if they disagree report it as noise rather than taking a third until they agree.

• A measurement is against whatever seed the app is running. make e2e needs SEED=default, so a confirming perf run taken straight after one measures a few dozen tracks: "Play 20 albums" becomes a dash and an artist's bytes fall 40×. Plausible in shape, meaningless. Restart on bulk before re-measuring anything.

• A longtask entry is delivered after the task that produced it. Reading window.__yjPerf.longtasks synchronously after the operation you just timed reports 0 ms of blocking beside a six-second stall. Wait a couple of hundred milliseconds first. The tell is that the two numbers in the row disagree — which is a good reason to always measure blocking and wall time.

• make dev-headless immediately after make sandbox-seed loses the race for port 34115 and comes up with no dev server, while still printing up. The measurement then attaches to a dying app. Sleep, or check curl -s -o /dev/null -w '%{http_code}' localhost:34115.

• search-bar debounces 150 ms. Anything measuring to the next frame measures the input echoing its own character.

• __yjEvents.wait() returns an already-buffered event. Without a reset() first you get the previous run's answer, which looks like a real result and is off by one iteration.

• A 0 ms result is usually a broken measurement, not a win. Waiting for #main-content > :not(.view-hidden) after a navigation matches the view being left — it stays on screen until the incoming one is ready — so every view reported 0 ms on every build. Wait for the specific element, never a generic selector. Same tell as the debounce: a number that cannot move is not evidence.

• git stash will not give you a baseline on a tree carrying uncommitted phases: stashing one file reverts every uncommitted change in it, not the one being measured. Build the before by undoing the single change by hand in the current file. For a cap or a threshold, setting the constant to Infinity is the cleanest possible one-variable undo.

• A bound cannot be verified by a run that never reaches it. The first bounded build measured identical to the unbounded one, because the session cached 180 entries against a cap of 192 and never evicted anything. Same tell as the two traps above — before and after suspiciously equal. Make the session overrun the limit.

• A negative result inherits the coverage of whatever produced it. Two sessions recorded the unbounded Explore caches as "does not reproduce" from a browse script that visits Explore and never searches in it — so both caches were empty the whole time. Before believing a finding did not reproduce, check the code path it names actually ran.

• A measurement that warms something has to run after everything that reads it. The playlist-open number pulls ~90 cover images; placed before the scroll measurement it filled the HTTP cache and took that row's request count from 26 to 0 — a clean, plausible, entirely fabricated improvement in a number nothing had touched. It runs last now, which costs it its own request count (zero on any build, so that row is in the JSON and off the table).

• The bulk library's covers are 300×300 and ~3.7 kB, deliberately (a realistic cover generator made a 2 GB library). Any finding about full-size artwork cannot show its magnitude here; measure the mechanism instead — e.g. which tier the request asked for rather than bytes saved.

Running the app

The app cannot be started without a display: devserver.Run ends in a blocking GTK window with no flag to suppress it. The harness gives it a virtual one and returns.

make sandbox-seed NAME=default    # once (~10 s; runs make testdata itself,
                                  # then builds a seed by running the app)
make dev-headless SEED=default    # starts in the background, returns when :34115 answers
make dev-logs                     # tail .dev/app.log
make dev-stop                     # SIGTERM, so shutdown hooks persist state

Then drive it. Run playwright-cli from the repo root — it picks up .playwright/cli.config.json from the cwd, and writes its snapshots and console logs to .playwright-cli/ relative to the cwd too. playwright-cli's own skill covers the commands; what is specific here is that a session must be named so it survives across separate shell calls:

playwright-cli -s=yj open http://localhost:34115
playwright-cli -s=yj snapshot                       # a11y tree, pierces shadow DOM
playwright-cli -s=yj screenshot --filename=/tmp/shot.png
playwright-cli -s=yj eval "() => window.__yjEvents.names()"
playwright-cli -s=yj eval "() => window.__yjEvents.call('queue.Queue.GetState', [], 5000)"
playwright-cli -s=yj click e391                     # ref from the snapshot
playwright-cli -s=yj close                          # `make dev-stop` does not do this

snapshot prints a path, not the tree — read the file it names, and check the timestamp, because a stale one from a previous session sits in the same directory.

.playwright/cli.config.json is picked up automatically: it sets the viewport, data-testid, and the init script that installs the event bridge. Assert on an event, not a timeout — half this app is push-driven. The bridge and the dev-only /__test/ control surface are documented in references/harness.md.

Other YJ_HOMEs exist for humans and block the terminal: make dev, make sandbox <name>, make fresh-install. Do not use them; you will never get the shell back.

Go, and the three build configurations

make test and make lint already run all three. Spell them out only when iterating on a single package:

go test ./backend/player/                        # the app build
go test -run TestName ./backend/player/
go test -tags indexbuild ./backend/explore/... ./cmd/...   # dump importer
go test -tags dev ./backend/testctl/...              # control surface

Forgetting the tag gives a build error that looks like a missing package. Audio integration tests additionally need YELLOWJACKET_INTEGRATION=1.

Three things golangci-lint v2 will reject that are easy to write: a dynamic fmt.Errorf without a sentinel (err113), a return with no blank line before it (nlreturn), and a long //nolint comment on the same line as its statement (golines reflows it and breaks the directive) — put the directive on its own line above.

Emit events through events.Emit(ctx, …), never runtime.EventsEmit. TestNoDirectRuntimeEmits walks the tree and fails the build otherwise, including in files no lint pass compiles.

References

• harness.md — the event bridge API, the /__test/ endpoints, and the config traps.
• fixtures.md — the generated library, the manifest, and selecting fixtures by case.
• ui-tier.md — how the Vitest tier fakes Wails, and what breaks in it.
• schema-change.md — the two-file schema/migration checklist.
• android-tier.md — the emulator tier, and the three reasons a failure there looks like a success. Read its first section before running anything on Android: Go's stdout does not reach logcat, os.Exit leaves no panic and no tombstone, and ActivityManager restarts a dying app fast enough that pidof always answers.