[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7EZ8AWERDMZBTPZSS3GSAEV":3,"skill-same:01M7EZ8AWERDMZBTPZSS3GSAEV":50},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":24,"commitSha":25,"skillDirectory":26,"forksCount":14,"repositoryHtmlUrl":27,"repositoryHomepageUrl":6,"ownerHtmlUrl":28,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":29,"install":30,"resources":43},"01M7EZ8AWERDMZBTPZSS3GSAEV","verify-omnigent",null,"Drive Omnigent the way a user does and prove a behavior with recorded evidence, using an isolated server, runner, and mock model plus a feature map of every user entry point. Load before reproducing a user-facing bug, before claiming a fix works, or when reviewing whether a change covered every surface (session vs. new-session composer, terminal strip vs. agent terminal, all twelve native harnesses). Covers the web UI, native harness terminals, and the CLI.","stephenhuo-code\u002Fcontexture-omnigent","stephenhuo-code","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F49703269?v=4","01M6CJ3KFPPGFV3Q04P3DAA687","yanwen","01M7EZ18M1YH045CYJR6EHFGM0",0,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Fstephenhuo-code\u002Fcontexture-omnigent\u002Fblob\u002Fmain\u002Ffeature-map\u002Fskills\u002Fverify-omnigent\u002FSKILL.md",[],[],"2026-11-16 10:05:34",false,"---\nname: verify-omnigent\ndescription: Drive Omnigent the way a user does and prove a behavior with recorded evidence, using an isolated server, runner, and mock model plus a feature map of every user entry point. Load before reproducing a user-facing bug, before claiming a fix works, or when reviewing whether a change covered every surface (session vs. new-session composer, terminal strip vs. agent terminal, all twelve native harnesses). Covers the web UI, native harness terminals, and the CLI.\n---\n\n# Verify Omnigent\n\nUse this skill to see a behavior happen in the real app and to prove a change,\nnot to reason about it from code. It has two parts:\n\n- **An isolated instance.** [`scripts\u002Fverify-env`](scripts\u002Fverify-env) wraps\n  `python -m dev.repro_env`: a server, runner, and mock model server on private\n  ports, with their own config, data, Claude, and Codex directories. It never\n  touches `~\u002F.omnigent`, a running host daemon, or another developer server.\n- **A feature map.** [feature map](..\u002F..\u002FREADME.md) lists each user-facing\n  feature's entry points, the tests that drive them, and the traps. A fix is\n  verified only when every entry point listed for its feature has proof.\n\nRun all commands from the repository root. Put `scripts\u002F` on your path or call\n`feature-map\u002Fskills\u002Fverify-omnigent\u002Fscripts\u002Fverify-env` directly.\n\n## Launch\n\n1. Install dependencies, Chromium, and build the web UI for the checkout:\n\n   ```sh\n   uv sync --frozen --extra all --group test\n   uv run --no-sync playwright install --with-deps chromium\n   pnpm install --frozen-lockfile --filter web && pnpm --filter web run build\n   ```\n\n2. Start an instance and load its paths:\n\n   ```sh\n   verify-env start          # waits until the runner is online, about 10 seconds\n   eval \"$(verify-env paths)\"\n   ```\n\n   `start` writes `VERIFY_ROOT` to `.omnigent\u002Fverify\u002Fcurrent`, so later shells\n   find the same instance. The instance stops itself after its lease (default\n   90 minutes; `--lease SECONDS`, 60 to 21600).\n\nReady means the server answers, the runner reports online, and the mock model\nserver answers. `start` fails with the environment's error and log path\notherwise.\n\nReady does not check browser launch. Use the Chromium version installed by\nthis checkout's Playwright package; if the environment provides it through\n`PLAYWRIGHT_BROWSERS_PATH`, keep that path available to the test process.\n`--ui-skip-build` reuses the existing bundle, so rebuild after frontend edits.\nNative-terminal journeys also need the relevant CLI and terminal prerequisites\n(such as tmux); a browser installation alone does not provide them.\n\nIn CI, the repro workflow already runs this environment. Use\n`python -m dev.repro_env exec -- ...` there instead of starting another.\n\n## Doctor\n\nRun `verify-env doctor` before the first drive, after any failed drive, and\nwhenever something looks off. It checks that the instance is ready, that its\nsupervisor is alive, that the runner and model server answer, and it warns when\nthe checkout has moved since launch. A warning about a moved checkout means the\ninstance runs old code: stop it and start a new one.\n\n## Drive\n\n1. Open the matching file in [feature map](..\u002F..\u002FREADME.md) and list every\n   entry point for the behavior in question.\n2. For each entry point, run the named test through the instance, recording on:\n\n   ```sh\n   verify-env run -- python -m pytest \u003Ctests\u002F...py::test_name> \\\n     --ui-skip-build --video=on --screenshot=on \\\n     --output=\"$VERIFY_EVIDENCE\u002F\u003Cfeature>\"\n   ```\n\n   Tests under `tests\u002Fbrowser_ui\u002F` need no instance:\n   `uv run pytest \u003Ctest> --browser-ui-skip-build --video=on --output=...`.\n   Tests the feature file marks \"own environment\" also run with plain\n   `uv run pytest`. Server\u002Ftransport\u002Fcomponent tests use plain pytest without\n   browser recording or build flags.\n3. For an entry point with no test, hand-drive it with Playwright against\n   `OMNIGENT_REPRO_SERVER_URL` inside `verify-env run -- python \u003Cscript>`, and\n   script model replies with the mock helpers named in the feature map README.\n4. To reproduce a bug, run the same journey on the unfixed code first and keep\n   that evidence; then run it again on the fix.\n\nA test that fails before reaching the user action has a setup failure, not a\nreproduction. Check its browser error and server\u002Frunner logs. When a test starts\nits own server or runner from inside an agent, check whether it inherited the\nparent's `OMNIGENT_RUNNER_*`, `RUNNER_SERVER_URL`, `OMNIGENT_PROCESS_LOG_FILE`,\nor `OMNIGENT_CONFIG_HOME`; a spawned host daemon writes its identity into that\nconfig home, so an inherited read-only path keeps it from coming online. Use an\nisolated child environment for that test;\ndo not change the controlling agent's environment. Record setup failures,\nskipped variants, and unavailable credentials separately from behavior results.\n\n## Evidence\n\nEverything under `$VERIFY_EVIDENCE` is the proof, one directory per feature.\nInstance logs, the database, and the mock model's recorded requests stay in\n`$VERIFY_ENV`. For each claim, record the checkout commit, installed harness versions,\nfeature file, entry point ID, command, and resulting artifact. Keep source\nreview, component tests, real process checks, browser drives, and live provider\nchecks distinct. A map contract pass verifies references and structure; it is\nnot evidence that the journeys ran. The proof standards are in the\n[feature map README](..\u002F..\u002FREADME.md#proof-and-coverage). Mock runs prove\nOmnigent's integration with Claude and Codex, not a live vendor model.\n\n## Cleanup\n\n`verify-env stop` stops only this instance's supervisor, which stops its own\nserver, runner, and model processes and saves the model's request log. It never\ndeletes `$VERIFY_ROOT`, so the evidence survives; remove old roots under\n`.omnigent\u002Fverify\u002F` or `\u002Ftmp\u002Fverify-omnigent.*` yourself once the evidence is\nattached. Never kill Omnigent processes by name: a developer's own server and\nhost daemon may be running on the same machine.\n\n## Helpers\n\n`scripts\u002Fverify-env` is the only helper. Its subcommands are `start [--lease\nSECONDS]`, `doctor`, `run -- COMMAND...`, `paths`, and `stop`; run it with no\narguments for usage.\n\nOn macOS the helper places the instance under `\u002Ftmp\u002Fverify-omnigent.*` when the\ncheckout path is long. macOS limits Unix socket paths to 104 bytes, and the\nenvironment's socket relay only works around long paths on Linux.\n\n## Keeping the map honest\n\nWhen a reviewer says a change missed a surface, add that surface to the feature\nfile in the same change. The checks and the weekly upkeep job are described in\n[Keeping the map current](..\u002F..\u002FREADME.md#keeping-the-map-current).\n","# Verify Omnigent\n\nUse this skill to see a behavior happen in the real app and to prove a change,\nnot to reason about it from code. It has two parts:\n\n- **An isolated instance.** [`scripts\u002Fverify-env`](scripts\u002Fverify-env) wraps\n  `python -m dev.repro_env`: a server, runner, and mock model server on private\n  ports, with their own config, data, Claude, and Codex directories. It never\n  touches `~\u002F.omnigent`, a running host daemon, or another developer server.\n- **A feature map.** [feature map](..\u002F..\u002FREADME.md) lists each user-facing\n  feature's entry points, the tests that drive them, and the traps. A fix is\n  verified only when every entry point listed for its feature has proof.\n\nRun all commands from the repository root. Put `scripts\u002F` on your path or call\n`feature-map\u002Fskills\u002Fverify-omnigent\u002Fscripts\u002Fverify-env` directly.\n\n## Launch\n\n1. Install dependencies, Chromium, and build the web UI for the checkout:\n\n   ```sh\n   uv sync --frozen --extra all --group test\n   uv run --no-sync playwright install --with-deps chromium\n   pnpm install --frozen-lockfile --filter web && pnpm --filter web run build\n   ```\n\n2. Start an instance and load its paths:\n\n   ```sh\n   verify-env start          # waits until the runner is online, about 10 seconds\n   eval \"$(verify-env paths)\"\n   ```\n\n   `start` writes `VERIFY_ROOT` to `.omnigent\u002Fverify\u002Fcurrent`, so later shells\n   find the same instance. The instance stops itself after its lease (default\n   90 minutes; `--lease SECONDS`, 60 to 21600).\n\nReady means the server answers, the runner reports online, and the mock model\nserver answers. `start` fails with the environment's error and log path\notherwise.\n\nReady does not check browser launch. Use the Chromium version installed by\nthis checkout's Playwright package; if the environment provides it through\n`PLAYWRIGHT_BROWSERS_PATH`, keep that path available to the test process.\n`--ui-skip-build` reuses the existing bundle, so rebuild after frontend edits.\nNative-terminal journeys also need the relevant CLI and terminal prerequisites\n(such as tmux); a browser installation alone does not provide them.\n\nIn CI, the repro workflow already runs this environment. Use\n`python -m dev.repro_env exec -- ...` there instead of starting another.\n\n## Doctor\n\nRun `verify-env doctor` before the first drive, after any failed drive, and\nwhenever something looks off. It checks that the instance is ready, that its\nsupervisor is alive, that the runner and model server answer, and it warns when\nthe checkout has moved since launch. A warning about a moved checkout means the\ninstance runs old code: stop it and start a new one.\n\n## Drive\n\n1. Open the matching file in [feature map](..\u002F..\u002FREADME.md) and list every\n   entry point for the behavior in question.\n2. For each entry point, run the named test through the instance, recording on:\n\n   ```sh\n   verify-env run -- python -m pytest \u003Ctests\u002F...py::test_name> \\\n     --ui-skip-build --video=on --screenshot=on \\\n     --output=\"$VERIFY_EVIDENCE\u002F\u003Cfeature>\"\n   ```\n\n   Tests under `tests\u002Fbrowser_ui\u002F` need no instance:\n   `uv run pytest \u003Ctest> --browser-ui-skip-build --video=on --output=...`.\n   Tests the feature file marks \"own environment\" also run with plain\n   `uv run pytest`. Server\u002Ftransport\u002Fcomponent tests use plain pytest without\n   browser recording or build flags.\n3. For an entry point with no test, hand-drive it with Playwright against\n   `OMNIGENT_REPRO_SERVER_URL` inside `verify-env run -- python \u003Cscript>`, and\n   script model replies with the mock helpers named in the feature map README.\n4. To reproduce a bug, run the same journey on the unfixed code first and keep\n   that evidence; then run it again on the fix.\n\nA test that fails before reaching the user action has a setup failure, not a\nreproduction. Check its browser error and server\u002Frunner logs. When a test starts\nits own server or runner from inside an agent, check whether it inherited the\nparent's `OMNIGENT_RUNNER_*`, `RUNNER_SERVER_URL`, `OMNIGENT_PROCESS_LOG_FILE`,\nor `OMNIGENT_CONFIG_HOME`; a spawned host daemon writes its identity into that\nconfig home, so an inherited read-only path keeps it from coming online. Use an\nisolated child environment for that test;\ndo not change the controlling agent's environment. Record setup failures,\nskipped variants, and unavailable credentials separately from behavior results.\n\n## Evidence\n\nEverything under `$VERIFY_EVIDENCE` is the proof, one directory per feature.\nInstance logs, the database, and the mock model's recorded requests stay in\n`$VERIFY_ENV`. For each claim, record the checkout commit, installed harness versions,\nfeature file, entry point ID, command, and resulting artifact. Keep source\nreview, component tests, real process checks, browser drives, and live provider\nchecks distinct. A map contract pass verifies references and structure; it is\nnot evidence that the journeys ran. The proof standards are in the\n[feature map README](..\u002F..\u002FREADME.md#proof-and-coverage). Mock runs prove\nOmnigent's integration with Claude and Codex, not a live vendor model.\n\n## Cleanup\n\n`verify-env stop` stops only this instance's supervisor, which stops its own\nserver, runner, and model processes and saves the model's request log. It never\ndeletes `$VERIFY_ROOT`, so the evidence survives; remove old roots under\n`.omnigent\u002Fverify\u002F` or `\u002Ftmp\u002Fverify-omnigent.*` yourself once the evidence is\nattached. Never kill Omnigent processes by name: a developer's own server and\nhost daemon may be running on the same machine.\n\n## Helpers\n\n`scripts\u002Fverify-env` is the only helper. Its subcommands are `start [--lease\nSECONDS]`, `doctor`, `run -- COMMAND...`, `paths`, and `stop`; run it with no\narguments for usage.\n\nOn macOS the helper places the instance under `\u002Ftmp\u002Fverify-omnigent.*` when the\ncheckout path is long. macOS limits Unix socket paths to 104 bytes, and the\nenvironment's socket relay only works around long paths on Linux.\n\n## Keeping the map honest\n\nWhen a reviewer says a change missed a surface, add that surface to the feature\nfile in the same change. The checks and the weekly upkeep job are described in\n[Keeping the map current](..\u002F..\u002FREADME.md#keeping-the-map-current).\n",{"name":5,"description":7},"main","1679c4238ee0448de3b26525da79184660132bf8","feature-map\u002Fskills\u002Fverify-omnigent","https:\u002F\u002Fgithub.com\u002Fstephenhuo-code\u002Fcontexture-omnigent","https:\u002F\u002Fgithub.com\u002Fstephenhuo-code","2026-11-09 19:15:17",{"source":8,"skillName":5,"commands":31,"prompt":41,"zipAvailable":42},[32,35,38],{"tool":33,"command":34},"npx","npx skills add stephenhuo-code\u002Fcontexture-omnigent --skill verify-omnigent",{"tool":36,"command":37},"bunx","bunx skills add stephenhuo-code\u002Fcontexture-omnigent --skill verify-omnigent",{"tool":39,"command":40},"pnpm","pnpm dlx skills add stephenhuo-code\u002Fcontexture-omnigent --skill verify-omnigent","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fstephenhuo-code\u002Fcontexture-omnigent\u002Fblob\u002Fmain\u002Ffeature-map\u002Fskills\u002Fverify-omnigent\u002FSKILL.md",true,[44],{"path":45,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":48,"sourceHtmlUrl":49},"scripts\u002Fverify-env","SCRIPT","脚本",4402,"https:\u002F\u002Fgithub.com\u002Fstephenhuo-code\u002Fcontexture-omnigent\u002Fblob\u002Fmain\u002Ffeature-map\u002Fskills\u002Fverify-omnigent\u002Fscripts\u002Fverify-env",[51,58,65,72,77,82],{"skillNo":52,"name":53,"nameZh":6,"description":54,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":55,"categories":56,"keywords":57,"publishedAt":19},"01M7EZ9NDK23Q45H6D3SW00YX6","deep-research","Research checklist and source-quality guidance for the archer e2e fixture.","https:\u002F\u002Fgithub.com\u002Fstephenhuo-code\u002Fcontexture-omnigent\u002Fblob\u002Fmain\u002Ftests\u002Fresources\u002Fexamples\u002Farcher\u002Fskills\u002Fdeep-research\u002FSKILL.md",[],[],{"skillNo":59,"name":60,"nameZh":6,"description":61,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":62,"categories":63,"keywords":64,"publishedAt":19},"01M7EZ9KFNY77TTBEBDJNEZTX8","pi-e2e-xyz-greet-c4a8d5","Test greeter — pi skills e2e. The unique suffix c4a8d5 appears in the skill name so a string match in the agent's output proves Pi loaded this skill.","https:\u002F\u002Fgithub.com\u002Fstephenhuo-code\u002Fcontexture-omnigent\u002Fblob\u002Fmain\u002Ftests\u002Fresources\u002Fagents\u002Fpi_skills_none\u002Fskills\u002Fpi-e2e-xyz-greet-c4a8d5\u002FSKILL.md",[],[],{"skillNo":66,"name":67,"nameZh":6,"description":68,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":69,"categories":70,"keywords":71,"publishedAt":19},"01M7EZ9H0BW3VKANPE96MA564A","pi-e2e-xyz-count-d2f6e1","Test counter — pi skills e2e. The unique suffix d2f6e1 appears in the skill name so a string match in the agent's output proves Pi loaded this skill.","https:\u002F\u002Fgithub.com\u002Fstephenhuo-code\u002Fcontexture-omnigent\u002Fblob\u002Fmain\u002Ftests\u002Fresources\u002Fagents\u002Fpi_skills_none\u002Fskills\u002Fpi-e2e-xyz-count-d2f6e1\u002FSKILL.md",[],[],{"skillNo":73,"name":60,"nameZh":6,"description":61,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":74,"categories":75,"keywords":76,"publishedAt":19},"01M7EZ9EHVAVZXA67B3WAD3D6T","https:\u002F\u002Fgithub.com\u002Fstephenhuo-code\u002Fcontexture-omnigent\u002Fblob\u002Fmain\u002Ftests\u002Fresources\u002Fagents\u002Fpi_skills_list\u002Fskills\u002Fpi-e2e-xyz-greet-c4a8d5\u002FSKILL.md",[],[],{"skillNo":78,"name":67,"nameZh":6,"description":68,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":79,"categories":80,"keywords":81,"publishedAt":19},"01M7EZ9C4CMPSQ8RYPFDCEEDBY","https:\u002F\u002Fgithub.com\u002Fstephenhuo-code\u002Fcontexture-omnigent\u002Fblob\u002Fmain\u002Ftests\u002Fresources\u002Fagents\u002Fpi_skills_list\u002Fskills\u002Fpi-e2e-xyz-count-d2f6e1\u002FSKILL.md",[],[],{"skillNo":83,"name":60,"nameZh":6,"description":61,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":84,"categories":85,"keywords":86,"publishedAt":19},"01M7EZ9A58NMY35T5SR2VQ5Z1V","https:\u002F\u002Fgithub.com\u002Fstephenhuo-code\u002Fcontexture-omnigent\u002Fblob\u002Fmain\u002Ftests\u002Fresources\u002Fagents\u002Fpi_skills_all\u002Fskills\u002Fpi-e2e-xyz-greet-c4a8d5\u002FSKILL.md",[],[]]