菜单
Run a single Node, browser, or Playwright spec instead of the full suite. Use when debugging one test, iterating on a failing spec, or the user asks to run one file, one test, or a filtered suite. Do not start with npm test.
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/LiberatedPixelCup/Universal-LPC-Spritesheet-Character-Generator --skill run-one-spec命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
Run and read unit-test coverage for this repo and satisfy the codecov/patch and codecov/changes gates. Use when a PR check fails on coverage, after editing gated production code under sources/ or scripts/, or when the user mentions coverage, Codecov, patch, or uncovered lines.
Thread CatalogReader and State through Mithril attrs, never a hidden global. Handle Result/LoadError, seed test catalogs, and keep URL hashes stable. Use when adding or editing a component, a browser spec, or sources/state/ code, or when the user mentions catalog, state, hash, or aliases.
Diagnose missing or stale dist/*-metadata.js, the Vite metadata plugin cache, and the ../name-metadata.js import alias. Use when dist/ is missing, a catalog resolves nothing, a spec fails only locally, or editing the Vite metadata plugin. Do not hand-edit dist/ or add another metadata module.
基于 SOC 职业分类
正在显示 SKILL.md
| name | run-one-spec |
| description | Run a single Node, browser, or Playwright spec instead of the full suite. Use when debugging one test, iterating on a failing spec, or the user asks to run one file, one test, or a filtered suite. Do not start with npm test. |
Pick the suite the file belongs to. Do not start with npm test.
Isolation is below. Coverage is a separate full-suite run — see
write-spec after a
gated sources/ or scripts/ edit (skip if the path is in
codecov.yml ignore:).
tests/node/)npm run test:node collects every *_spec.js / *_spec.ts and ignores extra
args. Call node --test on the file instead:
node --test tests/node/scripts/generateSources/items_spec.ts
node --test --test-name-pattern="parseItem" tests/node/scripts/generateSources/items_spec.ts
npm run test:node:coverage also runs every Node spec, not this file alone.
tests/tests.js imports every browser spec. There is
no CLI file filter. Do not comment out imports. A spec that is not imported
never runs; Codecov patch then fails with no Mocha failure.
Filter with ?grep= (best isolation). Start the server, then open the
harness with a grep pattern:
npm run test:server
# then open, in the launched browser:
# http://localhost:7357/tests_run.html?debug=false&grep=CategoryTree
Mocha's browser bundle reads grep, fgrep, and invert from
location.search inside mocha.run(), which
tests/tests.js calls. Every spec file is still
imported, so the filter selects which tests execute, not which files load.
The Mocha UI search box does the same thing by rewriting this query string.
If 7357 is busy, testem.cjs reads TESTEM_PORT:
TESTEM_PORT=7360 npm run test:server
it.only / describe.only works locally as a fallback. Do not commit it.
testem.cjs before_tests always runs every Node spec first, including this
Chrome-only command. It is still the full Mocha browser suite, not one file:
node ./node_modules/testem/testem.js ci --launch Chrome
npm run test:browser:coverage is that full suite in Chrome and Firefox.
DEBUG=1 (or DEBUG=true) keeps the app's verbose debug output instead of the
default ?debug=false.
npm run test:visual uses Playwright’s webServer (npm run dev locally).
npx playwright test with this repo’s config does the same.
npx playwright test tests/visual/home.spec.js
npx playwright test -g "title substring"
Wait for catalog/canvas via tests/visual/home-helpers.ts.
Do not register visual specs in tests/tests.js. They are not
Codecov-instrumented. Argos upload needs ARGOS_TOKEN. When a change
warrants a visual run: visual-test.