Verify
Build, launch and drive wayfinder-maps to verify a change end-to-end (server + headless browser).From its SKILL.md
npx -y skills add rengwu/wayfinder-maps --skill verifyAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
SKILL.md
2.9 KB, 744 tokens by cl100k_base, as published. Nobody here has run it
Verifying wayfinder-maps
Build & fixture
go build -o /tmp/wayfinder-maps ./cmd/wayfinder-maps
There is no sample map in the repo. Create an effort dir with map.md +
tickets/NN-slug.md. Ticket format: YAML frontmatter (type:, blocked_by: [01],
claimed_by:, undermined_by: [NN]) then # Title, ## Question; a ## Answer
section with prose = resolved, ## Ruled out = out_of_scope. Map format: # Name,
## Destination, and fog under ## Not yet specified as
- **Title.** clears-with: NN. Sanity-check the fixture with
wayfinder-maps status <effort-dir>.
Launch
PORT=78xx /tmp/wayfinder-maps serve <effort-dir> # pick a fresh port EVERY time
Gotcha: the user often has an instance (binary wm) already listening; a bind
failure only shows in the log while curl happily talks to the OLD server — check
the serve log says "serving", don't trust the port being answerable.
Dev mode: WAYFINDER_DEV=cmd/wayfinder-maps/web (path to web/ from cwd) serves
the frontend from disk with Cache-Control: no-store instead of the go:embed copy.
Drive (GUI surface)
Playwright's cached headless shell works without a full playwright install:
cd <scratch> && npm i playwright-core
# executablePath: ~/Library/Caches/ms-playwright/chromium_headless_shell-*/chrome-headless-shell-mac-arm64/chrome-headless-shell
- Wait ~4s after goto: the constellation fades in over 2.2s (a plain
--screenshot --virtual-time-budgetcapture shows HUD but NO stars — artifact, not a bug). - App state is inside ES modules (not on window). To click a specific star, find it by pixel colour on the canvas (frontier glow is gold) — layout is deterministic (fixed PRNG seed), so positions repeat across runs.
- Live-reload probe: append
## Answer\ntextto a ticket while the map is open; the 1.5s poller updates the HUD counts in ~2-3s. - Recents on the splash come from the user's real config — don't dismiss entries.
Flows worth driving: map render (all statuses + fog + edges), click star → panel
markdown, data-goto cross-ticket link, Escape closes, wheel zoom (labels thin
out), back button → splash/maplist.
WebKit
npx playwright-core install webkit (~77MB, cached in ms-playwright) then
require("playwright-core").webkit.launch() — no executablePath needed.
Covered headlessly: full map render, the label-alpha fade workaround (capture a
frame ~600ms into loadMap: labels must be dim WITH the stars, not opaque over
them), touch tap via a hasTouch context, wheel zoom, panel markdown.
NOT reachable headlessly — manual Safari checklist:
- Trackpad pinch zooms the map, not the page (gesture* events; Playwright's WebKit cannot construct GestureEvent, and Chromium never fires them).
- Touch pan/pinch on a real iOS device or simulator.
What ships with it
Read from the repository
Just SKILL.md. No reference files, no scripts.
Gives 0 of the 12 instructions most quality gates skills give in 744 tokens
Counted across 1,524 of the 2,830 authors here whose files we hold, read 2026-09-06
- Read full output and check exit codein 45 of 1524, across 40 files
- Verify output confirms the claimin 44 of 1524, across 39 files
- Identify the command that proves the claimin 43 of 1524, across 39 files
- Execute the full verification commandin 36 of 1524, across 30 files
- Produce a verification reportin 34 of 1524, across 18 files
- Review git diff changesin 30 of 1524, across 16 files
- Fix build failures immediatelyin 29 of 1524, across 9 files
- Group findings by severityin 28 of 1524
- State claim only with evidencein 27 of 1524, across 22 files
- Verify regression tests with red-green cyclein 26 of 1524, across 22 files
- Run the full test suitein 26 of 1524, across 25 files
- Run test suite with coveragein 25 of 1524, across 10 files
Said here and by no other author read
- Build the binary from source
- Create effort directory with map and ticket files
- Sanity check fixtures with status command
- Serve the map on a fresh port
- Verify server logs confirm active serving
- Wait four seconds after navigation for rendering
Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.