[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M78ZECT6QNHG9XCEJAF779V4":3,"skill-same:01M78ZECT6QNHG9XCEJAF779V4":45},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":15,"categories":16,"keywords":17,"publishedAt":18,"linkOnly":19,"markdownContent":20,"markdownBody":21,"frontMatter":22,"refName":23,"commitSha":24,"skillDirectory":25,"forksCount":26,"repositoryHtmlUrl":27,"repositoryHomepageUrl":28,"ownerHtmlUrl":29,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":30,"install":31,"resources":44},"01M78ZECT6QNHG9XCEJAF779V4","verify-project",null,"Verify an AKARI Video project with the L0 \u002F L1 \u002F L2 verification ladder required by an AKARI Video task contract on the current Theia stack. Read when acceptance criteria specify levels such as L0 or L0+L1, to determine exactly what to run and how at each level.（旧名: verify）","AkariLabs\u002Fakari-video","AkariLabs","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F309052672?v=4","01M78ZAVH5RSSCXA5RVB0RQYMW","01M78ZAVHS38JT84VMB4A3KFFB",214,"MIT","https:\u002F\u002Fgithub.com\u002FAkariLabs\u002Fakari-video\u002Fblob\u002Fmain\u002Fskills\u002Fverify-project\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: verify-project\ndescription: \"Verify an AKARI Video project with the L0 \u002F L1 \u002F L2 verification ladder required by an AKARI Video task contract on the current Theia stack. Read when acceptance criteria specify levels such as L0 or L0+L1, to determine exactly what to run and how at each level.（旧名: verify）\"\n---\n\n# verify — Verification ladder (L0 \u002F L1 \u002F L2)\n\n> **Language**: Respond in the user's language — conversation, questions, approval prompts, and the reports and records you write (including decision-log rows) follow the language the user writes in.\n\nThe mapping referenced by the task-contract template (internal `tasks\u002F_template\u002Ftask.md`).\n**These definitions are inferred from actual recent `report.md` \u002F `task.md` usage, not invented**;\nsee Sources at the end.\n\n## Mapping\n\n| Level | Definition | Judge | Examples on the current stack |\n|---|---|---|---|\n| L0 | Static\u002Fmechanical verification. Exit codes alone decide pass\u002Ffail; no real app launch or human viewing. | Agent alone | Minimum required by nearly every task |\n| L1 | Automated real-app observation. Launch Electron, operate via CDP\u002FPlaywright, capture screenshots, measure performance. Script-reproducible without subjective human judgment. | Agent alone (scripts) | Nearly all shell tasks of 2026-07-15\u002F16 require L0+L1 |\n| L2 | Judgments automation\u002Fscripts cannot fully make (subjective visual quality, owner's final check of actual distribution). | Human (owner) | **No examples on the current Theia stack**; see About L2 below |\n\n## L0 — Static and mechanical\n\nOn the current Theia stack, use `apps\u002Fshell` npm scripts:\n\n```sh\ncd apps\u002Fshell\nPYTHON=\u002Fusr\u002Fbin\u002Fpython3 npm ci --no-workspaces   # 初回・依存更新時のみ\nnpm run build:ext                                 # tsc -b（11 拡張の型検査 + コンパイル）\nnpm run lint                                       # eslint \"extensions\u002F*\u002Fsrc\u002F**\u002F*.{ts,tsx}\"\n```\n\n- **Use `npm ci`, not `npm install`.** `apps\u002Fshell\u002Fpackage-lock.json` has been **tracked** since 2026-08-19\n  (previously in `.gitignore`). Verification must not use `npm install`, which rewrites the lock against that day's\n  registry and introduces drift. Use `npm install` only for deliberate dependency upgrades, and commit the lock diff.\n- **`--no-workspaces`** treats apps\u002Fshell as standalone, using `apps\u002Fshell\u002Fpackage-lock.json` + `apps\u002Fshell\u002Fnode_modules`.\n  Without it, execution shifts to root workspace\u002Froot `package-lock.json`, involving `packages\u002F*` and roughly doubling runtime.\n- Install postinstall runs **`npm run preflight`** (`scripts\u002Fensure-no-nested-extension-deps.mjs` →\n  `scripts\u002Fensure-electron-dist.mjs`). `npm run build` prebuild and `npm start` prestart use the same preflight.\n  **No manual Electron extraction or nested `node_modules` cleanup is required**; both pitfalls below are repaired automatically.\n- Exit 0 from `build:ext` and `lint` is the minimum L0 pass. CI (`.github\u002Fworkflows\u002Fci.yml`) then runs apps\u002Fshell\n  and extension unit tests (`npm run test:shell`) on push\u002FPR, since 2026-09-02.\n  Installation uses `npm ci --no-workspaces --ignore-scripts` for the tracked lock.\n- **`packages\u002F*`, `scripts\u002Ftest`, and `skills\u002F*` unit tests are also in CI**, since 2026-09-02.\n  Canonical lane definitions: `scripts\u002Fci\u002Frun-unit-tests.mjs`.\n  `npm run test:unit` (pure, no external tools, required) \u002F `npm run test:quarantine` (already failing on main,\n  informational) \u002F `npm run test:media` (requires ffmpeg \u002F Chrome, informational) \u002F `npm run test:lanes`\n  (list and reasons for exclusions from CI). Locally, changed-package `npm test` suffices;\n  use `npm run test:unit` only to run the same set as CI.\n- Individual contracts may require stricter L0 `npm run build` (= `build:ext && theia build --mode production`),\n  including frontend\u002Fbackend\u002FElectron bundles (e.g., `shell-sb-surfaces` \u002F `shell-sc-project` \u002F `shell-sd-partner` \u002F\n  `governance-sanitize`). **Not run in CI**; see pitfalls below. Bundler is **esbuild**, not webpack\n  (Theia 1.73 default; absent `apps\u002Fshell\u002Fwebpack.config.js` means `theia build` always takes the esbuild route).\n- Tasks changing `packages\u002F*` use each package.json's build\u002Ftest scripts for L0\n  (`waveplan-2026-07-15-sbcd-e5.md`, acceptance-gate section).\n\n### Known pitfalls (L0)\n\n1. **npm 11's allow-scripts gate is opened through `allowScripts` in `apps\u002Fshell\u002Fpackage.json`**, an allowlist,\n   not a denylist. Measured 2026-08-19: after `npm ci --no-workspaces`, exactly **11 dependencies** have\n   install scripts (preinstall\u002Finstall\u002Fpostinstall + implicit `node-gyp rebuild` from `binding.gyp`), and **all 11\n   are listed in `allowScripts`, so they run**. Confirmed Electron postinstall via `npm rebuild electron --foreground-scripts`.\n   **The former statement that 11 packages' install scripts are skipped by default was wrong.** Entries use\n   `名前@完全一致バージョン` (existing machine format: name@exact-version), so **a single version mismatch silently\n   prevents that dependency's install script from running**. Tracked locks prevent drift; update `allowScripts`\n   with dependencies. `build:ext` \u002F `lint` need no native modules and pass either way.\n   CI L0 suppresses all scripts with `--ignore-scripts`.\n2. **Homebrew Python 3.14 and node-gyp**: On macOS, installation without `PYTHON=` can stop when\n   `drivelist`'s `node-gyp rebuild` hits Homebrew Python 3.14 `pyexpat` ABI mismatch\n   (`Symbol not found: _XML_SetAllocTrackerActivationThreshold`). Always set `PYTHON=\u002Fusr\u002Fbin\u002Fpython3` (Apple Python).\n3. **Electron postinstall silently exits partway on Node 26** (measured 2026-08-19).\n   `node_modules\u002Felectron\u002Finstall.js` extracts zips via extract-zip 2.0.1 → yauzl 2.10 → fd-slicer 1.1.\n   **Node v26.3.0 stops this pipe at 1,900,544 B (64KiB × 29 chunks); the event loop empties without `end`\n   or `error`, ending with exit 0.** `dist\u002F` contains only `LICENSE` and truncated `LICENSES.chromium.html`;\n   no `Electron.app`, `dist\u002Fversion`, or `path.txt`, yet installation appears successful.\n   Node 22.23.1 completes the same zip; standalone zlib\u002Ffd-slicer work on Node 26, indicating failure to resume reads\n   after backpressure clears. **`scripts\u002Fensure-electron-dist.mjs` preflight detects the incomplete result,\n   reextracts with `ditto` \u002F `tar` \u002F `unzip`, and writes `path.txt`.** Former manual `ditto -x -k` extraction\n   and `echo \u003Cversion> > dist\u002Fversion` workaround are no longer needed.\n4. **`apps\u002Fshell\u002Fpackage-lock.json` is tracked** (policy changed 2026-08-19; removed from `apps\u002Fshell\u002F.gitignore`).\n   Former claims that `npm ci` is unavailable or only Node pinning + `--no-workspaces` ensure reproducibility are **invalid**.\n   Before tracking, 63 of 1400 packages changed in 1 month (`@babel\u002F*`, acorn, postcss, react, terser,\n   webpack 5.108.4→5.109.2, etc.).\n5. **Zero grep matches means exit 1**. Add `|| true` when using grep in verification scripts.\n\n## L1 — Real-app observation (Electron via CDP\u002FPlaywright)\n\nNearly all current-stack tasks require L1 as mandatory GUI screenshots (`preview-streaming` \u002F `shell-s4-tabs` \u002F\n`preview-open-repair` \u002F `project-diff-repair` \u002F `shell-s12-preview-tab` \u002F `shell-sc-repair` \u002F\n`shell-strip-menu-repair`, etc.). Below is the reproducible procedure established by `shell-s4-tabs` report.md §6-2\nand `preview-streaming` report.md §2.\n\n**Never judge frames only at boundaries.** Exact beats\u002Fcut boundaries hit interval extremes (0% \u002F 100% = off-screen \u002F\nopacity 0), causing normal motion to be misdiagnosed. Always capture intermediate 1\u002F4, 1\u002F2, 3\u002F4 times too.\n`akari capture --auto` deterministically derives representative times including overlay\u002Fcaption midpoints.\n\n### Procedure\n\n1. **Build** (requires actual Electron in addition to L0):\n   ```sh\n   cd apps\u002Fshell\n   PYTHON=\u002Fusr\u002Fbin\u002Fpython3 npm ci --no-workspaces\n   npm run build\n   ```\n   Postinstall\u002Fprebuild `npm run preflight` supplies Electron (`node_modules\u002Felectron\u002Fdist` + `path.txt`),\n   fetching via `@electron\u002Fget` if no cached zip exists. **Manual `ditto` or `echo \u003Cversion> > dist\u002Fversion`\n   workarounds are no longer needed**. If necessary, preflight stops with exit 1 and a reason (L0 pitfall 3).\n2. **Prepare an isolated workspace**: Copy `templates\u002Fproject-default\u002F` to a temporary working directory,\n   leaving originals unchanged. For tasks needing media, generate actual files there with `ffmpeg` or similar.\n   Example: `preview-streaming` generated a 4K \u002F 120-second \u002F 836MB MP4 via `ffmpeg -f lavfi ...` for measurement.\n   Completely remove with `rm -rf` after checking; never commit it.\n3. **Launch Electron directly**, not `theia start` CLI or Playwright `_electron.launch()` (known argv bug in the latter):\n   ```sh\n   node_modules\u002Felectron\u002Fdist\u002FElectron.app\u002FContents\u002FMacOS\u002FElectron \\\n     \u003Capps\u002Fshell 絶対パス> \u003C隔離ワークスペース絶対パス> \\\n     --remote-debugging-port=\u003Cport> --user-data-dir=\u003C隔離ディレクトリ> --no-sandbox\n   ```\n   Pass environment `THEIA_CONFIG_DIR=\u003C隔離ディレクトリ>` (existing isolated-directory placeholder)\n   to avoid conflicts with normal development settings.\n4. **`npm install` `playwright-core` only in a verification scratch directory**, never in the repository;\n   attach with `chromium.connectOverCDP('http:\u002F\u002F127.0.0.1:\u003Cport>')`.\n5. **Reach nested webview frames**: Theia `WebviewWidget` nests an outer `webview.localhost` iframe around\n   inner `active-frame`. Simple `page.frames()` may not reach actual content. A custom client may need to correlate\n   CDP `Page.getFrameTree` with `Runtime.executionContextCreated` `auxData.frameId`, identify the actual inner\n   execution context, then read DOM state (`readyState` \u002F `currentTime` \u002F `data-*` attributes) via `Runtime.evaluate`.\n   See `preview-streaming` report.md §2.\n6. **Observe and record**:\n   - Screenshots: CDP `Page.captureScreenshot` or Playwright screenshot API.\n   - **Do not commit observation records**, since 2026-10-02. Screenshots, video, audio, logs, measurement JSON,\n     etc., belong outside the repository (temporary working directories or task-record storage); reports contain\n     summaries\u002Fnumbers. Only reproduction scripts (`.mjs` \u002F `.cjs` \u002F `.js` \u002F `.sh`) and `README.md` may be committed\n     beneath `evidence\u002F\u003C機能名>\u002F` (existing feature-name placeholder). Governance `evidence-growth` rejects new\n     records under `evidence\u002F` (images\u002Fvideo\u002Faudio\u002F`.txt`\u002F`.log`\u002F`.json`, etc.). Keep already committed evidence unchanged.\n   - Performance: Sample `ps -o %cpu -p \u003Cpid>` every 1 second, recording GPU\u002Frenderer\u002Fmain separately.\n     Measure timing through CDP events\u002FDOM changes before\u002Fafter operations, e.g., milliseconds to `readyState>=3`.\n   - API boundaries: Directly call services through Inversify DI via `window.theia.container` to verify backend\n     rejection even when frontend guards are bypassed (defense in depth). See `preview-streaming` report's serving-boundary checks.\n7. **Cleanup**: Identify actual launched Electron PIDs with `ps aux` and `kill` those PIDs;\n   never use broad matching such as `pkill -f`. Completely delete isolated workspaces\u002Flarge generated media afterward;\n   never commit them.\n\n### Known pitfalls (L1)\n\n0. **Residual `ELECTRON_RUN_AS_NODE=1` immediately crashes Electron** (measured 2026-09-27,\n   `library-tile-cards` lane). Electron starts as Node rather than GUI, leaving `app` undefined and failing with:\n   ```\n   Cannot read properties of undefined (reading 'requestSingleInstanceLock')\n   ```\n   This occurs if L1 follows packaged-path verification (`ELECTRON_RUN_AS_NODE=1` + `cwd=\u002F`, L0 pitfall 3)\n   in the same shell. Always remove it at launch with `env -u ELECTRON_RUN_AS_NODE \u003Celectron のパス> …`\n   (existing Electron-path placeholder).\n\n1. **Nested extension `node_modules` (the most frequent L1 pitfall)**: A second `@theia\u002Fcore` inside\n   `apps\u002Fshell\u002Fextensions\u002F\u003Cext>\u002Fnode_modules\u002F` causes esbuild's realpath-based bundling to include 2\n   `frontend-application-config-provider` copies in one `bundle.js`. `src-gen\u002Ffrontend\u002Findex.js` calls `set()`\n   on one, while extensions call `get()` on the other, leaving the frontend at its preload spinner with:\n   ```\n   Failed to start the frontend application.\n   Error: The configuration is not set. Did you call FrontendApplicationConfigProvider#set?\n   ```\n   This dual-package hazard was **reproduced and confirmed on a real app 2026-08-19**: the same tree errors only\n   when `extensions\u002Fakari-preview\u002Fnode_modules\u002F@theia\u002Fcore` exists; deleting it restores launch.\n   Cause: standalone extension-directory `npm install` (one run installs 643 packages).\n   **Never run `npm install` inside extension directories.** Preflight (postinstall\u002Fprebuild\u002Fprestart) removes\n   accidental nesting, but `bundle.js` needs rebuilding; rerun from `npm run build`.\n   A prior case blamed webpack versions (2026-08-18 timeline-z-order-unification handoff §2).\n   `theia build` uses esbuild and never calls webpack; launch was measured with webpack 5.109.2 still installed.\n   Suspect nested `node_modules` first.\n2. **Electron entry point**: Use `lib\u002Fbackend\u002Felectron-main.js` as specified by `package.json`'s `main`.\n   Directly specifying `lib\u002Fbackend\u002Fmain.js` creates no `BrowserWindow`, leaving `\u002Fjson\u002Flist` empty.\n3. **Packaged path resolution**: Bundled `__dirname` differs from development (e.g., `app.asar\u002Flib\u002Fbackend`).\n   Features searching relative paths outside `apps\u002Fshell` must be checked against `electron-builder --dir` output\n   with `ELECTRON_RUN_AS_NODE=1` + `cwd=\u002F` (see `package-runtime-assets` task.md).\n   Checking only development `npm start` is insufficient.\n\n## About L2\n\n**No recent tasks after migration to the current Theia stack** (2026-07-15\u002F16, S-A–S-D, S4\u002FS6\u002FS12,\npreview-streaming, preview-open-repair, project-diff-repair, package-runtime-assets, shell-sc-repair,\nshell-strip-menu-repair, etc.) **have required L2 as an acceptance criterion**.\nAll finish within L0 + L1 (mandatory real-app GUI screenshots); automated real-app observation (L1)\nis the practical highest level.\n\nL2 was used by verify during the old Tauri era. Examples informing the definition:\n\n- `2026-07-15-preview-band-artifact`: L2 reproduced and verified the fix on a real app, with **the owner's own\n  visual confirmation** that the band disappeared as the final gate.\n- `2026-07-15-vlog-mvp-edit`: L2 real-app check unperformed. L1 actual export\u002Fffprobe verification was complete,\n  but real-app visual verification of preview compositing was separately classified L2 and outside scope.\n- `2026-07-15-decision-cards-runtime`: Real Tauri\u002Fcompanion-extension webview visual inspection was outside scope\n  (L2 excluded); automated headless Chrome DOM checks (L1 equivalent) were distinct from actual webview experience (L2).\n- Private dev-harness contract (2026-07-14) explicitly equates its L3 real-device layer with verify L2;\n  L3 means owner field test → handoff record (private wave plan, 2026-07-15, §6).\n\nThe shared distinction is **L2 means judgments automation\u002Fscripts cannot fully make**, not whether real media is used.\nL1 commonly uses real media too, such as `preview-streaming`'s 836MB 4K footage.\nThus define current-stack L2 as follows, explicitly noting this is extrapolation without current-stack examples:\n\n> L2: The owner personally checks real-use scenarios on packaged distribution (`electron-builder` output),\n> assessing visual feel\u002Fusability and other aspects automation cannot judge, and records the result in handoff.\n\n### Criteria for requiring L2\n\n- Acceptance includes **subjective quality judgments** (e.g., “feels natural”, “looks faithful”), not numerically measurable.\n- A needed experience cannot in principle be reproduced by L1 CDP\u002FPlaywright observation\n  (actual distribution install → launch → operation, real touch\u002Fmouse feel, etc.).\n- The coordinator\u002Fowner explicitly requests a real-device field test.\n\nOtherwise finish current-stack tasks at L0+L1, matching actual practice.\n\n## Which levels to require (task-side criteria)\n\n- **L0 only**: No observable runtime behavior change (documentation, CI configuration, type definitions, etc.).\n  This task itself (`ci-and-verify-skill`) is an example.\n- **L0+L1**: Default for code changes affecting GUI\u002Fruntime, such as `apps\u002Fshell\u002Fextensions\u002F**` or\n  `packages\u002Fpreview-engine`. Nearly all 2026-07-15\u002F16 shell tasks belong here.\n- **L0+L1+L2**: Only when the L2 criteria above apply. No current-stack precedent.\n\n## Sources (basis for inference)\n\nMapping\u002Fprocedures are inferred from private internal task records (`akari-video-internal`) report.md\u002Ftask.md.\nThose internal records do not belong in this repository, consistent with\n`https:\u002F\u002Fgithub.com\u002FAkariLabs\u002Fakari-video\u002Fblob\u002Fmain\u002Fdocs\u002Fdesign-2026-07-13-agent-native-architecture.md` §3. Executable verification procedures (commands, observation\nsubjects, pitfalls) are already provided in L0\u002FL1 above. Below only identifies source tasks, all private:\n\n- preview-streaming (2026-07-16): L1 reproduction, 836MB real media, nested-webview access.\n- shell-s4-tabs (2026-07-16): L1 reproduction, dual-package hazard, Electron entry-point pitfalls; procedures in §6-2.\n- package-runtime-assets (2026-07-16): Packaged L1 execution.\n- Internal task-record README: CI prerequisites for acceptance flow 4-c.\n- Internal task template: Source of the verification-level mapping terminology.\n- dev-harness contract (2026-07-14), wave plan (2026-07-15) §6: Verification layer table, L0\u002FL1\u002FL2\u002FL3 definitions.\n- preview-band-artifact, vlog-mvp-edit, decision-cards-runtime (all 2026-07-15, old Tauri era):\n  L2 examples; these alone support the L2 definition because no current-stack examples exist.\n","# verify — Verification ladder (L0 \u002F L1 \u002F L2)\n\n> **Language**: Respond in the user's language — conversation, questions, approval prompts, and the reports and records you write (including decision-log rows) follow the language the user writes in.\n\nThe mapping referenced by the task-contract template (internal `tasks\u002F_template\u002Ftask.md`).\n**These definitions are inferred from actual recent `report.md` \u002F `task.md` usage, not invented**;\nsee Sources at the end.\n\n## Mapping\n\n| Level | Definition | Judge | Examples on the current stack |\n|---|---|---|---|\n| L0 | Static\u002Fmechanical verification. Exit codes alone decide pass\u002Ffail; no real app launch or human viewing. | Agent alone | Minimum required by nearly every task |\n| L1 | Automated real-app observation. Launch Electron, operate via CDP\u002FPlaywright, capture screenshots, measure performance. Script-reproducible without subjective human judgment. | Agent alone (scripts) | Nearly all shell tasks of 2026-07-15\u002F16 require L0+L1 |\n| L2 | Judgments automation\u002Fscripts cannot fully make (subjective visual quality, owner's final check of actual distribution). | Human (owner) | **No examples on the current Theia stack**; see About L2 below |\n\n## L0 — Static and mechanical\n\nOn the current Theia stack, use `apps\u002Fshell` npm scripts:\n\n```sh\ncd apps\u002Fshell\nPYTHON=\u002Fusr\u002Fbin\u002Fpython3 npm ci --no-workspaces   # 初回・依存更新時のみ\nnpm run build:ext                                 # tsc -b（11 拡張の型検査 + コンパイル）\nnpm run lint                                       # eslint \"extensions\u002F*\u002Fsrc\u002F**\u002F*.{ts,tsx}\"\n```\n\n- **Use `npm ci`, not `npm install`.** `apps\u002Fshell\u002Fpackage-lock.json` has been **tracked** since 2026-08-19\n  (previously in `.gitignore`). Verification must not use `npm install`, which rewrites the lock against that day's\n  registry and introduces drift. Use `npm install` only for deliberate dependency upgrades, and commit the lock diff.\n- **`--no-workspaces`** treats apps\u002Fshell as standalone, using `apps\u002Fshell\u002Fpackage-lock.json` + `apps\u002Fshell\u002Fnode_modules`.\n  Without it, execution shifts to root workspace\u002Froot `package-lock.json`, involving `packages\u002F*` and roughly doubling runtime.\n- Install postinstall runs **`npm run preflight`** (`scripts\u002Fensure-no-nested-extension-deps.mjs` →\n  `scripts\u002Fensure-electron-dist.mjs`). `npm run build` prebuild and `npm start` prestart use the same preflight.\n  **No manual Electron extraction or nested `node_modules` cleanup is required**; both pitfalls below are repaired automatically.\n- Exit 0 from `build:ext` and `lint` is the minimum L0 pass. CI (`.github\u002Fworkflows\u002Fci.yml`) then runs apps\u002Fshell\n  and extension unit tests (`npm run test:shell`) on push\u002FPR, since 2026-09-02.\n  Installation uses `npm ci --no-workspaces --ignore-scripts` for the tracked lock.\n- **`packages\u002F*`, `scripts\u002Ftest`, and `skills\u002F*` unit tests are also in CI**, since 2026-09-02.\n  Canonical lane definitions: `scripts\u002Fci\u002Frun-unit-tests.mjs`.\n  `npm run test:unit` (pure, no external tools, required) \u002F `npm run test:quarantine` (already failing on main,\n  informational) \u002F `npm run test:media` (requires ffmpeg \u002F Chrome, informational) \u002F `npm run test:lanes`\n  (list and reasons for exclusions from CI). Locally, changed-package `npm test` suffices;\n  use `npm run test:unit` only to run the same set as CI.\n- Individual contracts may require stricter L0 `npm run build` (= `build:ext && theia build --mode production`),\n  including frontend\u002Fbackend\u002FElectron bundles (e.g., `shell-sb-surfaces` \u002F `shell-sc-project` \u002F `shell-sd-partner` \u002F\n  `governance-sanitize`). **Not run in CI**; see pitfalls below. Bundler is **esbuild**, not webpack\n  (Theia 1.73 default; absent `apps\u002Fshell\u002Fwebpack.config.js` means `theia build` always takes the esbuild route).\n- Tasks changing `packages\u002F*` use each package.json's build\u002Ftest scripts for L0\n  (`waveplan-2026-07-15-sbcd-e5.md`, acceptance-gate section).\n\n### Known pitfalls (L0)\n\n1. **npm 11's allow-scripts gate is opened through `allowScripts` in `apps\u002Fshell\u002Fpackage.json`**, an allowlist,\n   not a denylist. Measured 2026-08-19: after `npm ci --no-workspaces`, exactly **11 dependencies** have\n   install scripts (preinstall\u002Finstall\u002Fpostinstall + implicit `node-gyp rebuild` from `binding.gyp`), and **all 11\n   are listed in `allowScripts`, so they run**. Confirmed Electron postinstall via `npm rebuild electron --foreground-scripts`.\n   **The former statement that 11 packages' install scripts are skipped by default was wrong.** Entries use\n   `名前@完全一致バージョン` (existing machine format: name@exact-version), so **a single version mismatch silently\n   prevents that dependency's install script from running**. Tracked locks prevent drift; update `allowScripts`\n   with dependencies. `build:ext` \u002F `lint` need no native modules and pass either way.\n   CI L0 suppresses all scripts with `--ignore-scripts`.\n2. **Homebrew Python 3.14 and node-gyp**: On macOS, installation without `PYTHON=` can stop when\n   `drivelist`'s `node-gyp rebuild` hits Homebrew Python 3.14 `pyexpat` ABI mismatch\n   (`Symbol not found: _XML_SetAllocTrackerActivationThreshold`). Always set `PYTHON=\u002Fusr\u002Fbin\u002Fpython3` (Apple Python).\n3. **Electron postinstall silently exits partway on Node 26** (measured 2026-08-19).\n   `node_modules\u002Felectron\u002Finstall.js` extracts zips via extract-zip 2.0.1 → yauzl 2.10 → fd-slicer 1.1.\n   **Node v26.3.0 stops this pipe at 1,900,544 B (64KiB × 29 chunks); the event loop empties without `end`\n   or `error`, ending with exit 0.** `dist\u002F` contains only `LICENSE` and truncated `LICENSES.chromium.html`;\n   no `Electron.app`, `dist\u002Fversion`, or `path.txt`, yet installation appears successful.\n   Node 22.23.1 completes the same zip; standalone zlib\u002Ffd-slicer work on Node 26, indicating failure to resume reads\n   after backpressure clears. **`scripts\u002Fensure-electron-dist.mjs` preflight detects the incomplete result,\n   reextracts with `ditto` \u002F `tar` \u002F `unzip`, and writes `path.txt`.** Former manual `ditto -x -k` extraction\n   and `echo \u003Cversion> > dist\u002Fversion` workaround are no longer needed.\n4. **`apps\u002Fshell\u002Fpackage-lock.json` is tracked** (policy changed 2026-08-19; removed from `apps\u002Fshell\u002F.gitignore`).\n   Former claims that `npm ci` is unavailable or only Node pinning + `--no-workspaces` ensure reproducibility are **invalid**.\n   Before tracking, 63 of 1400 packages changed in 1 month (`@babel\u002F*`, acorn, postcss, react, terser,\n   webpack 5.108.4→5.109.2, etc.).\n5. **Zero grep matches means exit 1**. Add `|| true` when using grep in verification scripts.\n\n## L1 — Real-app observation (Electron via CDP\u002FPlaywright)\n\nNearly all current-stack tasks require L1 as mandatory GUI screenshots (`preview-streaming` \u002F `shell-s4-tabs` \u002F\n`preview-open-repair` \u002F `project-diff-repair` \u002F `shell-s12-preview-tab` \u002F `shell-sc-repair` \u002F\n`shell-strip-menu-repair`, etc.). Below is the reproducible procedure established by `shell-s4-tabs` report.md §6-2\nand `preview-streaming` report.md §2.\n\n**Never judge frames only at boundaries.** Exact beats\u002Fcut boundaries hit interval extremes (0% \u002F 100% = off-screen \u002F\nopacity 0), causing normal motion to be misdiagnosed. Always capture intermediate 1\u002F4, 1\u002F2, 3\u002F4 times too.\n`akari capture --auto` deterministically derives representative times including overlay\u002Fcaption midpoints.\n\n### Procedure\n\n1. **Build** (requires actual Electron in addition to L0):\n   ```sh\n   cd apps\u002Fshell\n   PYTHON=\u002Fusr\u002Fbin\u002Fpython3 npm ci --no-workspaces\n   npm run build\n   ```\n   Postinstall\u002Fprebuild `npm run preflight` supplies Electron (`node_modules\u002Felectron\u002Fdist` + `path.txt`),\n   fetching via `@electron\u002Fget` if no cached zip exists. **Manual `ditto` or `echo \u003Cversion> > dist\u002Fversion`\n   workarounds are no longer needed**. If necessary, preflight stops with exit 1 and a reason (L0 pitfall 3).\n2. **Prepare an isolated workspace**: Copy `templates\u002Fproject-default\u002F` to a temporary working directory,\n   leaving originals unchanged. For tasks needing media, generate actual files there with `ffmpeg` or similar.\n   Example: `preview-streaming` generated a 4K \u002F 120-second \u002F 836MB MP4 via `ffmpeg -f lavfi ...` for measurement.\n   Completely remove with `rm -rf` after checking; never commit it.\n3. **Launch Electron directly**, not `theia start` CLI or Playwright `_electron.launch()` (known argv bug in the latter):\n   ```sh\n   node_modules\u002Felectron\u002Fdist\u002FElectron.app\u002FContents\u002FMacOS\u002FElectron \\\n     \u003Capps\u002Fshell 絶対パス> \u003C隔離ワークスペース絶対パス> \\\n     --remote-debugging-port=\u003Cport> --user-data-dir=\u003C隔離ディレクトリ> --no-sandbox\n   ```\n   Pass environment `THEIA_CONFIG_DIR=\u003C隔離ディレクトリ>` (existing isolated-directory placeholder)\n   to avoid conflicts with normal development settings.\n4. **`npm install` `playwright-core` only in a verification scratch directory**, never in the repository;\n   attach with `chromium.connectOverCDP('http:\u002F\u002F127.0.0.1:\u003Cport>')`.\n5. **Reach nested webview frames**: Theia `WebviewWidget` nests an outer `webview.localhost` iframe around\n   inner `active-frame`. Simple `page.frames()` may not reach actual content. A custom client may need to correlate\n   CDP `Page.getFrameTree` with `Runtime.executionContextCreated` `auxData.frameId`, identify the actual inner\n   execution context, then read DOM state (`readyState` \u002F `currentTime` \u002F `data-*` attributes) via `Runtime.evaluate`.\n   See `preview-streaming` report.md §2.\n6. **Observe and record**:\n   - Screenshots: CDP `Page.captureScreenshot` or Playwright screenshot API.\n   - **Do not commit observation records**, since 2026-10-02. Screenshots, video, audio, logs, measurement JSON,\n     etc., belong outside the repository (temporary working directories or task-record storage); reports contain\n     summaries\u002Fnumbers. Only reproduction scripts (`.mjs` \u002F `.cjs` \u002F `.js` \u002F `.sh`) and `README.md` may be committed\n     beneath `evidence\u002F\u003C機能名>\u002F` (existing feature-name placeholder). Governance `evidence-growth` rejects new\n     records under `evidence\u002F` (images\u002Fvideo\u002Faudio\u002F`.txt`\u002F`.log`\u002F`.json`, etc.). Keep already committed evidence unchanged.\n   - Performance: Sample `ps -o %cpu -p \u003Cpid>` every 1 second, recording GPU\u002Frenderer\u002Fmain separately.\n     Measure timing through CDP events\u002FDOM changes before\u002Fafter operations, e.g., milliseconds to `readyState>=3`.\n   - API boundaries: Directly call services through Inversify DI via `window.theia.container` to verify backend\n     rejection even when frontend guards are bypassed (defense in depth). See `preview-streaming` report's serving-boundary checks.\n7. **Cleanup**: Identify actual launched Electron PIDs with `ps aux` and `kill` those PIDs;\n   never use broad matching such as `pkill -f`. Completely delete isolated workspaces\u002Flarge generated media afterward;\n   never commit them.\n\n### Known pitfalls (L1)\n\n0. **Residual `ELECTRON_RUN_AS_NODE=1` immediately crashes Electron** (measured 2026-09-27,\n   `library-tile-cards` lane). Electron starts as Node rather than GUI, leaving `app` undefined and failing with:\n   ```\n   Cannot read properties of undefined (reading 'requestSingleInstanceLock')\n   ```\n   This occurs if L1 follows packaged-path verification (`ELECTRON_RUN_AS_NODE=1` + `cwd=\u002F`, L0 pitfall 3)\n   in the same shell. Always remove it at launch with `env -u ELECTRON_RUN_AS_NODE \u003Celectron のパス> …`\n   (existing Electron-path placeholder).\n\n1. **Nested extension `node_modules` (the most frequent L1 pitfall)**: A second `@theia\u002Fcore` inside\n   `apps\u002Fshell\u002Fextensions\u002F\u003Cext>\u002Fnode_modules\u002F` causes esbuild's realpath-based bundling to include 2\n   `frontend-application-config-provider` copies in one `bundle.js`. `src-gen\u002Ffrontend\u002Findex.js` calls `set()`\n   on one, while extensions call `get()` on the other, leaving the frontend at its preload spinner with:\n   ```\n   Failed to start the frontend application.\n   Error: The configuration is not set. Did you call FrontendApplicationConfigProvider#set?\n   ```\n   This dual-package hazard was **reproduced and confirmed on a real app 2026-08-19**: the same tree errors only\n   when `extensions\u002Fakari-preview\u002Fnode_modules\u002F@theia\u002Fcore` exists; deleting it restores launch.\n   Cause: standalone extension-directory `npm install` (one run installs 643 packages).\n   **Never run `npm install` inside extension directories.** Preflight (postinstall\u002Fprebuild\u002Fprestart) removes\n   accidental nesting, but `bundle.js` needs rebuilding; rerun from `npm run build`.\n   A prior case blamed webpack versions (2026-08-18 timeline-z-order-unification handoff §2).\n   `theia build` uses esbuild and never calls webpack; launch was measured with webpack 5.109.2 still installed.\n   Suspect nested `node_modules` first.\n2. **Electron entry point**: Use `lib\u002Fbackend\u002Felectron-main.js` as specified by `package.json`'s `main`.\n   Directly specifying `lib\u002Fbackend\u002Fmain.js` creates no `BrowserWindow`, leaving `\u002Fjson\u002Flist` empty.\n3. **Packaged path resolution**: Bundled `__dirname` differs from development (e.g., `app.asar\u002Flib\u002Fbackend`).\n   Features searching relative paths outside `apps\u002Fshell` must be checked against `electron-builder --dir` output\n   with `ELECTRON_RUN_AS_NODE=1` + `cwd=\u002F` (see `package-runtime-assets` task.md).\n   Checking only development `npm start` is insufficient.\n\n## About L2\n\n**No recent tasks after migration to the current Theia stack** (2026-07-15\u002F16, S-A–S-D, S4\u002FS6\u002FS12,\npreview-streaming, preview-open-repair, project-diff-repair, package-runtime-assets, shell-sc-repair,\nshell-strip-menu-repair, etc.) **have required L2 as an acceptance criterion**.\nAll finish within L0 + L1 (mandatory real-app GUI screenshots); automated real-app observation (L1)\nis the practical highest level.\n\nL2 was used by verify during the old Tauri era. Examples informing the definition:\n\n- `2026-07-15-preview-band-artifact`: L2 reproduced and verified the fix on a real app, with **the owner's own\n  visual confirmation** that the band disappeared as the final gate.\n- `2026-07-15-vlog-mvp-edit`: L2 real-app check unperformed. L1 actual export\u002Fffprobe verification was complete,\n  but real-app visual verification of preview compositing was separately classified L2 and outside scope.\n- `2026-07-15-decision-cards-runtime`: Real Tauri\u002Fcompanion-extension webview visual inspection was outside scope\n  (L2 excluded); automated headless Chrome DOM checks (L1 equivalent) were distinct from actual webview experience (L2).\n- Private dev-harness contract (2026-07-14) explicitly equates its L3 real-device layer with verify L2;\n  L3 means owner field test → handoff record (private wave plan, 2026-07-15, §6).\n\nThe shared distinction is **L2 means judgments automation\u002Fscripts cannot fully make**, not whether real media is used.\nL1 commonly uses real media too, such as `preview-streaming`'s 836MB 4K footage.\nThus define current-stack L2 as follows, explicitly noting this is extrapolation without current-stack examples:\n\n> L2: The owner personally checks real-use scenarios on packaged distribution (`electron-builder` output),\n> assessing visual feel\u002Fusability and other aspects automation cannot judge, and records the result in handoff.\n\n### Criteria for requiring L2\n\n- Acceptance includes **subjective quality judgments** (e.g., “feels natural”, “looks faithful”), not numerically measurable.\n- A needed experience cannot in principle be reproduced by L1 CDP\u002FPlaywright observation\n  (actual distribution install → launch → operation, real touch\u002Fmouse feel, etc.).\n- The coordinator\u002Fowner explicitly requests a real-device field test.\n\nOtherwise finish current-stack tasks at L0+L1, matching actual practice.\n\n## Which levels to require (task-side criteria)\n\n- **L0 only**: No observable runtime behavior change (documentation, CI configuration, type definitions, etc.).\n  This task itself (`ci-and-verify-skill`) is an example.\n- **L0+L1**: Default for code changes affecting GUI\u002Fruntime, such as `apps\u002Fshell\u002Fextensions\u002F**` or\n  `packages\u002Fpreview-engine`. Nearly all 2026-07-15\u002F16 shell tasks belong here.\n- **L0+L1+L2**: Only when the L2 criteria above apply. No current-stack precedent.\n\n## Sources (basis for inference)\n\nMapping\u002Fprocedures are inferred from private internal task records (`akari-video-internal`) report.md\u002Ftask.md.\nThose internal records do not belong in this repository, consistent with\n`https:\u002F\u002Fgithub.com\u002FAkariLabs\u002Fakari-video\u002Fblob\u002Fmain\u002Fdocs\u002Fdesign-2026-07-13-agent-native-architecture.md` §3. Executable verification procedures (commands, observation\nsubjects, pitfalls) are already provided in L0\u002FL1 above. Below only identifies source tasks, all private:\n\n- preview-streaming (2026-07-16): L1 reproduction, 836MB real media, nested-webview access.\n- shell-s4-tabs (2026-07-16): L1 reproduction, dual-package hazard, Electron entry-point pitfalls; procedures in §6-2.\n- package-runtime-assets (2026-07-16): Packaged L1 execution.\n- Internal task-record README: CI prerequisites for acceptance flow 4-c.\n- Internal task template: Source of the verification-level mapping terminology.\n- dev-harness contract (2026-07-14), wave plan (2026-07-15) §6: Verification layer table, L0\u002FL1\u002FL2\u002FL3 definitions.\n- preview-band-artifact, vlog-mvp-edit, decision-cards-runtime (all 2026-07-15, old Tauri era):\n  L2 examples; these alone support the L2 definition because no current-stack examples exist.\n",{"name":5,"description":7},"main","12b63130cc05dfc1efff973508a33f4fe3a27129","skills\u002Fverify-project",25,"https:\u002F\u002Fgithub.com\u002FAkariLabs\u002Fakari-video","https:\u002F\u002Fakari.video","https:\u002F\u002Fgithub.com\u002FAkariLabs","2026-10-22 01:34:40",{"source":8,"skillName":5,"commands":32,"prompt":42,"zipAvailable":43},[33,36,39],{"tool":34,"command":35},"npx","npx skills add AkariLabs\u002Fakari-video --skill verify-project",{"tool":37,"command":38},"bunx","bunx skills add AkariLabs\u002Fakari-video --skill verify-project",{"tool":40,"command":41},"pnpm","pnpm dlx skills add AkariLabs\u002Fakari-video --skill verify-project","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FAkariLabs\u002Fakari-video\u002Fblob\u002Fmain\u002Fskills\u002Fverify-project\u002FSKILL.md",true,[],[46,53,60,67,74,81],{"skillNo":47,"name":48,"nameZh":6,"description":49,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":50,"categories":51,"keywords":52,"publishedAt":18},"01M78ZE9H36DBKGW1DH7PRZ526","verify-edit","Verify edit.json and optional analysis.json \u002F captions.json \u002F media with a deterministic CLI, then finish QA with frame inspection and reporting after PASS. Use immediately after writing or modifying edit.json, before export, and after applying review feedback.（旧名: edit-lint）","https:\u002F\u002Fgithub.com\u002FAkariLabs\u002Fakari-video\u002Fblob\u002Fmain\u002Fskills\u002Fverify-edit\u002FSKILL.md",[],[],{"skillNo":54,"name":55,"nameZh":6,"description":56,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":57,"categories":58,"keywords":59,"publishedAt":18},"01M78ZE5CYGEAKZG6M5SEP2B20","setup-remote","Set up remote access to approval reports and previews from another device such as a phone, and send footage to the workspace. Use doctor to assess Tailscale, guide installation and login (human actions), expose the preview server (default 4567) and approval report helper over tailnet-only HTTPS with tailscale serve, connect Taildrop reception to workspace inbox\u002F, and verify access from another device. Use for requests to view reports on a phone, approve remotely, send footage from a phone, or set up remote access. Public internet exposure (funnel) is outside the default scope.","https:\u002F\u002Fgithub.com\u002FAkariLabs\u002Fakari-video\u002Fblob\u002Fmain\u002Fskills\u002Fsetup-remote\u002FSKILL.md",[],[],{"skillNo":61,"name":62,"nameZh":6,"description":63,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":64,"categories":65,"keywords":66,"publishedAt":18},"01M78ZE2HZR93E1ZJ5GPG2V0K5","setup-project","Set up an AKARI Video project headlessly. Recursively copy `templates\u002Fproject-default\u002F`, record the template version, initialize git only when safe, and generate an HTML creation report. No app launch is needed. Use to create a video project or fill in an existing folder as an AKARI Video project.（旧名: create-project）","https:\u002F\u002Fgithub.com\u002FAkariLabs\u002Fakari-video\u002Fblob\u002Fmain\u002Fskills\u002Fsetup-project\u002FSKILL.md",[],[],{"skillNo":68,"name":69,"nameZh":6,"description":70,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":71,"categories":72,"keywords":73,"publishedAt":18},"01M78ZDWRXJJF2B98WGQ35W53E","setup-library","Use for first-time AKARI Video setup or when the current project needs more usable assets. This first-run skill checks ffmpeg \u002F whisper-cli (and can install with approval), reads the bundled catalog with akari assets to propose a starter pack, obtains explicit human approval, then acquires, places, validates, and updates INDEX entries. It also covers the audio library (BGM and sound effects): bulk acquisition of first-party AKARI Sounds from GitHub Releases, a free-source candidate HTML list with drop-folder matching or assisted acquisition for missing families, and keep\u002Fdrop decisions in the listening gallery. Use for requests such as \"I want more BGM\" or \"add sound effects to the library\".（旧名: setup-audio-library）","https:\u002F\u002Fgithub.com\u002FAkariLabs\u002Fakari-video\u002Fblob\u002Fmain\u002Fskills\u002Fsetup-library\u002FSKILL.md",[],[],{"skillNo":75,"name":76,"nameZh":6,"description":77,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":78,"categories":79,"keywords":80,"publishedAt":18},"01M78ZDNSC34WR5NYJYNZ4HC28","setup-connections","Set up and inspect AKARI Video connections: AI-service keys, which models are allowed, cost-approval policy, the tool and media-service state, and 3D settings. Use for first-time setup, adding a provider or key later, checking what is usable (「何が使える？」 'what can I use?'), setting a key such as fal (「fal の鍵を設定したい」), 3D settings (「3D の設定を使いたい」), connection checks, and gates before paid generation or external publication. Model selection belongs to the production skills (they read the ledger and preferences); this skill decides allowed models and the cost policy. Works with `.akari\u002Fconnections.json` and a free, read-only doctor.（旧名: manage-connections）（旧名: connect）","https:\u002F\u002Fgithub.com\u002FAkariLabs\u002Fakari-video\u002Fblob\u002Fmain\u002Fskills\u002Fsetup-connections\u002FSKILL.md",[],[],{"skillNo":82,"name":83,"nameZh":6,"description":84,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":85,"categories":86,"keywords":87,"publishedAt":18},"01M78ZDGV662PQ1259SCVW9XHN","setup-chat-approval","Set up Telegram notifications at approval gates and return approvals with a button tap. Use doctor to assess state, guide BotFather token issuance and credentials.env registration (human actions), identify the chat ID, register the connection, and verify real-device notification and button approval. Use for requests to approve in chat, receive phone notifications, or configure a Telegram bot. Free-text instructions and agent startup are outside scope (separate contract).","https:\u002F\u002Fgithub.com\u002FAkariLabs\u002Fakari-video\u002Fblob\u002Fmain\u002Fskills\u002Fsetup-chat-approval\u002FSKILL.md",[],[]]