[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M79FBYHBWMNRPAS3E1R1AF98":3,"skill-same:01M79FBYHBWMNRPAS3E1R1AF98":67},{"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":27,"commitSha":28,"skillDirectory":29,"forksCount":14,"repositoryHtmlUrl":30,"repositoryHomepageUrl":6,"ownerHtmlUrl":31,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":32,"install":33,"resources":46},"01M79FBYHBWMNRPAS3E1R1AF98","mcore-testing",null,"Test system for Megatron-LM. Covers test layout, recipe YAML structure, adding and running unit and functional tests, golden values, marker filters, and CI parity.","Elogic360\u002Fsoftware-factory","Elogic360","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F156535519?v=4","01M74NH3F0RB13PWB7JZDE2V80","elogic360","01M74NH3FPC65082MDZVMFHDM3",0,"Apache-2.0","https:\u002F\u002Fgithub.com\u002FElogic360\u002Fsoftware-factory\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Fmcore-testing\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: mcore-testing\ndescription: Test system for Megatron-LM. Covers test layout, recipe YAML structure, adding and running unit and functional tests, golden values, marker filters, and CI parity.\nlicense: Apache-2.0\nwhen_to_use: Adding or running a unit or functional test; understanding the test layout; writing a recipe YAML; downloading or updating golden values; reproducing a test failure locally; 'how do I add a test', 'run unit tests', 'pytest fails', 'test layout', 'golden values', 'recipe YAML', 'marker filter'.\nmetadata:\n  author: Philip Petrakian \u003Cppetrakian@nvidia.com>\n---\n\n# Testing Guide\n\n---\n\n## Answer-First Testing Facts\n\nFor questions about disabling tests without deleting them:\n\n- Functional recipe entries stay in YAML; disable by suffixing scope with\n  `-broken`, for example `scope: [mr-github]` -> `scope: [mr-github-broken]`.\n- Unit-test skips use pytest markers instead: `@pytest.mark.flaky_in_dev` skips\n  in the default dev environment, and `@pytest.mark.flaky` skips in LTS.\n- Do not delete the test case or recipe entry when the goal is discoverability\n  and easy re-enable.\n\n---\n\n## Test Layout\n\n```text\ntests\u002F\n├── unit_tests\u002F          # pytest, 1 node × 8 GPUs, torch.distributed runner\n├── functional_tests\u002F    # end-to-end shell + training scripts\n│   └── test_cases\u002F\n│       └── {model}\u002F{test_case}\u002F\n│           ├── model_config.yaml          # training args\n│           └── golden_values_{env}_{platform}.json\n└── test_utils\u002F\n    ├── recipes\u002F\n    │   ├── h100\u002F        # YAML recipes for H100 jobs\n    │   └── gb200\u002F       # YAML recipes for GB200 jobs\n    └── python_scripts\u002F  # helpers (recipe_parser, golden-value download, …)\n```\n\n---\n\n## How Tests Execute\n\nThe GitHub Actions runner invokes `launch_nemo_run_workload.py`, which uses\n**nemo-run** to launch a `DockerExecutor` container. The repo is bind-mounted\nat `\u002Fopt\u002Fmegatron-lm`; training data is mounted at `\u002Fmnt\u002Fartifacts`.\n\n**Unit tests** are dispatched through `torch.distributed.run`:\n\n- Ranks 0 and 3 are tee-d to stdout; all other ranks write only to log files.\n- Per-rank log files land at `{assets_dir}\u002Flogs\u002F1\u002F` and are uploaded as a\n  GitHub artifact after the run.\n\n**Functional tests** are driven by\n`tests\u002Ffunctional_tests\u002Fshell_test_utils\u002Frun_ci_test.sh`. Only rank 0 runs the\npytest validation step; training output from all ranks is uploaded as an artifact.\n\n**Flaky-failure auto-retry**: `launch_nemo_run_workload.py` retries up to\n**3 times** for known transient patterns (NCCL timeout, ECC error, segfault,\nHuggingFace connectivity, …) before declaring a genuine failure.\n\n---\n\n## Recipe YAML Structure\n\nRecipes live in `tests\u002Ftest_utils\u002Frecipes\u002F` and are parsed by\n`tests\u002Ftest_utils\u002Fpython_scripts\u002Frecipe_parser.py`. Each file expands a\ncartesian `products` block into individual workload specs:\n\n```yaml\ntype: basic\nformat_version: 1\nmaintainers: [mcore]\nloggers: [stdout]\nspec:\n  name: \"{test_case}_{environment}_{platforms}\"\n  model: gpt              # maps to tests\u002Ffunctional_tests\u002Ftest_cases\u002F{model}\u002F\n  build: mcore-pyt-{environment}\n  nodes: 1\n  gpus: 8\n  n_repeat: 5\n  platforms: dgx_h100\n  time_limit: 1800\n  script_setup: |\n    ...\n  script: |-\n    bash tests\u002Ffunctional_tests\u002Fshell_test_utils\u002Frun_ci_test.sh ...\nproducts:\n  - test_case: [my_test]\n    products:\n      - environment: [dev, lts]\n        scope: [mr-github]\n        platforms: [dgx_h100]\n```\n\nKey runtime placeholders: `{assets_dir}`, `{artifacts_dir}`, `{test_case}`,\n`{environment}`, `{platforms}`, `{n_repeat}`.\n\n### Disabling a Test Without Deleting It\n\nTo temporarily disable a test case in a recipe YAML, suffix its `scope` value\nwith `-broken` — **do not delete the entry**:\n\n```yaml\n# before (test runs in CI)\nscope: [mr-github]\n\n# after (test is skipped; entry preserved for easy re-enable)\nscope: [mr-github-broken]\n```\n\n---\n\n## Running Unit Tests Locally\n\nAll unit tests initialize a `torch.distributed` group, so every invocation\nrequires GPU access and must go through `torch.distributed.run`:\n\n```bash\n# Full suite\nuv run python -m torch.distributed.run --nproc-per-node 8 -m pytest -q \\\n  tests\u002Funit_tests\n\n# Single file\nuv run python -m torch.distributed.run --nproc-per-node 8 -m pytest -q \\\n  tests\u002Funit_tests\u002Fmodels\u002Ftest_gpt_model.py\n\n# Single test\nuv run python -m torch.distributed.run --nproc-per-node 8 -m pytest -q \\\n  tests\u002Funit_tests\u002Fmodels\u002Ftest_gpt_model.py::TestGPTModel::test_constructor\n\n# Filter by name substring\nuv run python -m torch.distributed.run --nproc-per-node 8 -m pytest -q \\\n  tests\u002Funit_tests -k optimizer\n```\n\n### Marker filters\n\n```bash\n# Exclude flaky tests during development\nuv run python -m torch.distributed.run --nproc-per-node 8 -m pytest -q \\\n  tests\u002Funit_tests -m \"not flaky and not flaky_in_dev\"\n\n# Include experimental tests\nuv run python -m torch.distributed.run --nproc-per-node 8 -m pytest -q \\\n  tests\u002Funit_tests --experimental\n```\n\n### CI parity\n\nUse `tests\u002Funit_tests\u002Frun_ci_test.sh` to reproduce a CI bucket failure exactly.\nFor ad-hoc runs, prefer the direct `torch.distributed.run` invocations above.\n\n### Gotchas\n\n- `pyproject.toml` sets `addopts = --durations=15 -s -rA` — stdout is not\n  captured (`-s`), so ranks interleave during multi-rank runs. Override with\n  `--capture=fd` when debugging a specific rank.\n- `tests\u002Funit_tests\u002Fconftest.py` looks for test data under `\u002Fopt\u002Fdata` and\n  attempts a download if missing. Supply it manually or skip data-dependent\n  tests when running outside the canonical container.\n\n---\n\n## Adding a Unit Test\n\n1. Create `tests\u002Funit_tests\u002F\u003Ccategory>\u002Ftest_\u003Cname>.py`.\n2. Use fixtures from `tests\u002Funit_tests\u002Fconftest.py`.\n3. Apply markers as needed:\n   - `@pytest.mark.internal` — skipped on `legacy` tag\n   - `@pytest.mark.flaky_in_dev` — skipped in `dev` environment (CI default; use this to disable a flaky test without blocking the standard pipeline)\n   - `@pytest.mark.flaky` — skipped in `lts` environment\n   - `@pytest.mark.experimental` — `latest` tag only\n4. Verify locally (see Running Unit Tests Locally above).\n5. If the test needs a dedicated CI bucket, add an entry to\n   `tests\u002Ftest_utils\u002Frecipes\u002Fh100\u002Funit-tests.yaml`.\n\n---\n\n## Adding a Functional \u002F Integration Test\n\n1. Create `tests\u002Ffunctional_tests\u002Ftest_cases\u002F\u003Cmodel>\u002F\u003Ctest_name>\u002F`.\n2. Write `model_config.yaml` with `MODEL_ARGS`, `ENV_VARS`, and `TEST_TYPE`.\n3. Add a YAML recipe under `tests\u002Ftest_utils\u002Frecipes\u002Fh100\u002F` (and `gb200\u002F` if\n   needed). Required fields: `scope`, `environment`, `platform`, `n_repeat`,\n   `time_limit`.\n4. Push the PR, add the label **\"Run functional tests\"** to trigger a full run.\n5. After a successful run, download golden values:\n\n   ```bash\n   python tests\u002Ftest_utils\u002Fpython_scripts\u002Fdownload_golden_values.py \\\n     --source github --pipeline-id \u003Crun-id>\n   ```\n\n6. Commit the downloaded golden values.\n\n---\n\n## Common Pitfalls\n\n| Problem | Cause | Fix |\n|---------|-------|-----|\n| Test passes locally but fails in CI | Different environment or data path | Check `DATA_PATH`, `DATA_CACHE_PATH`, and the `environment` tag (`dev` vs `lts`) |\n| Golden value mismatch after a code change | Numerical regression | Download new golden values via `download_golden_values.py` after a clean run |\n| `cicd-integration-tests-gb200` not triggered | GB200 jobs require maintainer status | Ask a maintainer to trigger, or add the `Run functional tests` label |\n","# Testing Guide\n\n---\n\n## Answer-First Testing Facts\n\nFor questions about disabling tests without deleting them:\n\n- Functional recipe entries stay in YAML; disable by suffixing scope with\n  `-broken`, for example `scope: [mr-github]` -> `scope: [mr-github-broken]`.\n- Unit-test skips use pytest markers instead: `@pytest.mark.flaky_in_dev` skips\n  in the default dev environment, and `@pytest.mark.flaky` skips in LTS.\n- Do not delete the test case or recipe entry when the goal is discoverability\n  and easy re-enable.\n\n---\n\n## Test Layout\n\n```text\ntests\u002F\n├── unit_tests\u002F          # pytest, 1 node × 8 GPUs, torch.distributed runner\n├── functional_tests\u002F    # end-to-end shell + training scripts\n│   └── test_cases\u002F\n│       └── {model}\u002F{test_case}\u002F\n│           ├── model_config.yaml          # training args\n│           └── golden_values_{env}_{platform}.json\n└── test_utils\u002F\n    ├── recipes\u002F\n    │   ├── h100\u002F        # YAML recipes for H100 jobs\n    │   └── gb200\u002F       # YAML recipes for GB200 jobs\n    └── python_scripts\u002F  # helpers (recipe_parser, golden-value download, …)\n```\n\n---\n\n## How Tests Execute\n\nThe GitHub Actions runner invokes `launch_nemo_run_workload.py`, which uses\n**nemo-run** to launch a `DockerExecutor` container. The repo is bind-mounted\nat `\u002Fopt\u002Fmegatron-lm`; training data is mounted at `\u002Fmnt\u002Fartifacts`.\n\n**Unit tests** are dispatched through `torch.distributed.run`:\n\n- Ranks 0 and 3 are tee-d to stdout; all other ranks write only to log files.\n- Per-rank log files land at `{assets_dir}\u002Flogs\u002F1\u002F` and are uploaded as a\n  GitHub artifact after the run.\n\n**Functional tests** are driven by\n`tests\u002Ffunctional_tests\u002Fshell_test_utils\u002Frun_ci_test.sh`. Only rank 0 runs the\npytest validation step; training output from all ranks is uploaded as an artifact.\n\n**Flaky-failure auto-retry**: `launch_nemo_run_workload.py` retries up to\n**3 times** for known transient patterns (NCCL timeout, ECC error, segfault,\nHuggingFace connectivity, …) before declaring a genuine failure.\n\n---\n\n## Recipe YAML Structure\n\nRecipes live in `tests\u002Ftest_utils\u002Frecipes\u002F` and are parsed by\n`tests\u002Ftest_utils\u002Fpython_scripts\u002Frecipe_parser.py`. Each file expands a\ncartesian `products` block into individual workload specs:\n\n```yaml\ntype: basic\nformat_version: 1\nmaintainers: [mcore]\nloggers: [stdout]\nspec:\n  name: \"{test_case}_{environment}_{platforms}\"\n  model: gpt              # maps to tests\u002Ffunctional_tests\u002Ftest_cases\u002F{model}\u002F\n  build: mcore-pyt-{environment}\n  nodes: 1\n  gpus: 8\n  n_repeat: 5\n  platforms: dgx_h100\n  time_limit: 1800\n  script_setup: |\n    ...\n  script: |-\n    bash tests\u002Ffunctional_tests\u002Fshell_test_utils\u002Frun_ci_test.sh ...\nproducts:\n  - test_case: [my_test]\n    products:\n      - environment: [dev, lts]\n        scope: [mr-github]\n        platforms: [dgx_h100]\n```\n\nKey runtime placeholders: `{assets_dir}`, `{artifacts_dir}`, `{test_case}`,\n`{environment}`, `{platforms}`, `{n_repeat}`.\n\n### Disabling a Test Without Deleting It\n\nTo temporarily disable a test case in a recipe YAML, suffix its `scope` value\nwith `-broken` — **do not delete the entry**:\n\n```yaml\n# before (test runs in CI)\nscope: [mr-github]\n\n# after (test is skipped; entry preserved for easy re-enable)\nscope: [mr-github-broken]\n```\n\n---\n\n## Running Unit Tests Locally\n\nAll unit tests initialize a `torch.distributed` group, so every invocation\nrequires GPU access and must go through `torch.distributed.run`:\n\n```bash\n# Full suite\nuv run python -m torch.distributed.run --nproc-per-node 8 -m pytest -q \\\n  tests\u002Funit_tests\n\n# Single file\nuv run python -m torch.distributed.run --nproc-per-node 8 -m pytest -q \\\n  tests\u002Funit_tests\u002Fmodels\u002Ftest_gpt_model.py\n\n# Single test\nuv run python -m torch.distributed.run --nproc-per-node 8 -m pytest -q \\\n  tests\u002Funit_tests\u002Fmodels\u002Ftest_gpt_model.py::TestGPTModel::test_constructor\n\n# Filter by name substring\nuv run python -m torch.distributed.run --nproc-per-node 8 -m pytest -q \\\n  tests\u002Funit_tests -k optimizer\n```\n\n### Marker filters\n\n```bash\n# Exclude flaky tests during development\nuv run python -m torch.distributed.run --nproc-per-node 8 -m pytest -q \\\n  tests\u002Funit_tests -m \"not flaky and not flaky_in_dev\"\n\n# Include experimental tests\nuv run python -m torch.distributed.run --nproc-per-node 8 -m pytest -q \\\n  tests\u002Funit_tests --experimental\n```\n\n### CI parity\n\nUse `tests\u002Funit_tests\u002Frun_ci_test.sh` to reproduce a CI bucket failure exactly.\nFor ad-hoc runs, prefer the direct `torch.distributed.run` invocations above.\n\n### Gotchas\n\n- `pyproject.toml` sets `addopts = --durations=15 -s -rA` — stdout is not\n  captured (`-s`), so ranks interleave during multi-rank runs. Override with\n  `--capture=fd` when debugging a specific rank.\n- `tests\u002Funit_tests\u002Fconftest.py` looks for test data under `\u002Fopt\u002Fdata` and\n  attempts a download if missing. Supply it manually or skip data-dependent\n  tests when running outside the canonical container.\n\n---\n\n## Adding a Unit Test\n\n1. Create `tests\u002Funit_tests\u002F\u003Ccategory>\u002Ftest_\u003Cname>.py`.\n2. Use fixtures from `tests\u002Funit_tests\u002Fconftest.py`.\n3. Apply markers as needed:\n   - `@pytest.mark.internal` — skipped on `legacy` tag\n   - `@pytest.mark.flaky_in_dev` — skipped in `dev` environment (CI default; use this to disable a flaky test without blocking the standard pipeline)\n   - `@pytest.mark.flaky` — skipped in `lts` environment\n   - `@pytest.mark.experimental` — `latest` tag only\n4. Verify locally (see Running Unit Tests Locally above).\n5. If the test needs a dedicated CI bucket, add an entry to\n   `tests\u002Ftest_utils\u002Frecipes\u002Fh100\u002Funit-tests.yaml`.\n\n---\n\n## Adding a Functional \u002F Integration Test\n\n1. Create `tests\u002Ffunctional_tests\u002Ftest_cases\u002F\u003Cmodel>\u002F\u003Ctest_name>\u002F`.\n2. Write `model_config.yaml` with `MODEL_ARGS`, `ENV_VARS`, and `TEST_TYPE`.\n3. Add a YAML recipe under `tests\u002Ftest_utils\u002Frecipes\u002Fh100\u002F` (and `gb200\u002F` if\n   needed). Required fields: `scope`, `environment`, `platform`, `n_repeat`,\n   `time_limit`.\n4. Push the PR, add the label **\"Run functional tests\"** to trigger a full run.\n5. After a successful run, download golden values:\n\n   ```bash\n   python tests\u002Ftest_utils\u002Fpython_scripts\u002Fdownload_golden_values.py \\\n     --source github --pipeline-id \u003Crun-id>\n   ```\n\n6. Commit the downloaded golden values.\n\n---\n\n## Common Pitfalls\n\n| Problem | Cause | Fix |\n|---------|-------|-----|\n| Test passes locally but fails in CI | Different environment or data path | Check `DATA_PATH`, `DATA_CACHE_PATH`, and the `environment` tag (`dev` vs `lts`) |\n| Golden value mismatch after a code change | Numerical regression | Download new golden values via `download_golden_values.py` after a clean run |\n| `cicd-integration-tests-gb200` not triggered | GB200 jobs require maintainer status | Ask a maintainer to trigger, or add the `Run functional tests` label |\n",{"name":5,"license":15,"metadata":24,"description":7,"when_to_use":26},{"author":25},"Philip Petrakian \u003Cppetrakian@nvidia.com>","Adding or running a unit or functional test; understanding the test layout; writing a recipe YAML; downloading or updating golden values; reproducing a test failure locally; 'how do I add a test', 'run unit tests', 'pytest fails', 'test layout', 'golden values', 'recipe YAML', 'marker filter'.","main","c475fc64e94e803b50d14a791dfee8853fbb2c5c","skills\u002Fnvidia\u002Fmcore-testing","https:\u002F\u002Fgithub.com\u002FElogic360\u002Fsoftware-factory","https:\u002F\u002Fgithub.com\u002FElogic360","2026-11-08 22:21:22",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":45},[35,38,41],{"tool":36,"command":37},"npx","npx skills add Elogic360\u002Fsoftware-factory --skill mcore-testing",{"tool":39,"command":40},"bunx","bunx skills add Elogic360\u002Fsoftware-factory --skill mcore-testing",{"tool":42,"command":43},"pnpm","pnpm dlx skills add Elogic360\u002Fsoftware-factory --skill mcore-testing","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FElogic360\u002Fsoftware-factory\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Fmcore-testing\u002FSKILL.md",true,[47,53,57,61],{"path":48,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":51,"sourceHtmlUrl":52},"BENCHMARK.md","REFERENCE","参考文档",2747,"https:\u002F\u002Fgithub.com\u002FElogic360\u002Fsoftware-factory\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Fmcore-testing\u002FBENCHMARK.md",{"path":54,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":55,"sourceHtmlUrl":56},"evals\u002Fevals.json",3,"https:\u002F\u002Fgithub.com\u002FElogic360\u002Fsoftware-factory\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Fmcore-testing\u002Fevals\u002Fevals.json",{"path":58,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":59,"sourceHtmlUrl":60},"skill-card.md",2533,"https:\u002F\u002Fgithub.com\u002FElogic360\u002Fsoftware-factory\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Fmcore-testing\u002Fskill-card.md",{"path":62,"resourceType":63,"resourceTypeDesc":64,"sizeBytes":65,"sourceHtmlUrl":66},"skill.oms.sig","OTHER","其他",4585,"https:\u002F\u002Fgithub.com\u002FElogic360\u002Fsoftware-factory\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Fmcore-testing\u002Fskill.oms.sig",[68,75,82,89,96,103],{"skillNo":69,"name":70,"nameZh":6,"description":71,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":72,"categories":73,"keywords":74,"publishedAt":19},"01M79G206MT3J6S0GY3C45P2KV","vss-summarize-video","Use to summarize a recorded video via the LVS summarization microservice (HITL-gated) with a VLM fallback. Not for report generation or live RTSP captioning.","https:\u002F\u002Fgithub.com\u002FElogic360\u002Fsoftware-factory\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Fvss-summarize-video\u002FSKILL.md",[],[],{"skillNo":76,"name":77,"nameZh":6,"description":78,"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},"01M79G1WN63PGRE8CKD47WEBQ5","vss-setup-video-analytics-api","Use to deploy the vss-video-analytics-api REST service standalone (config-source, data-log bind, Elasticsearch, optional Kafka). Not for full warehouse deploy.","https:\u002F\u002Fgithub.com\u002FElogic360\u002Fsoftware-factory\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Fvss-setup-video-analytics-api\u002FSKILL.md",[],[],{"skillNo":83,"name":84,"nameZh":6,"description":85,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":86,"categories":87,"keywords":88,"publishedAt":19},"01M79G1S9F7VY3EF6H53PRP5AN","vss-setup-behavior-analytics","Use to deploy the vss-behavior-analytics service standalone (entrypoint, config-source, optional calibration). Not for the full warehouse deploy.","https:\u002F\u002Fgithub.com\u002FElogic360\u002Fsoftware-factory\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Fvss-setup-behavior-analytics\u002FSKILL.md",[],[],{"skillNo":90,"name":91,"nameZh":6,"description":92,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":93,"categories":94,"keywords":95,"publishedAt":19},"01M79G1N1S1Y27PFZPYMNK1MTQ","vss-search-archive","Use this skill to run top-level VSS fusion search on archived video, or to ingest video files \u002F RTSP streams for search. Do NOT use for ad-hoc visual Q&A (use vss-ask-video), live captioning (use vss-deploy-dense-captioning), or video summarization and reports (use vss-summarize-video).","https:\u002F\u002Fgithub.com\u002FElogic360\u002Fsoftware-factory\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Fvss-search-archive\u002FSKILL.md",[],[],{"skillNo":97,"name":98,"nameZh":6,"description":99,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":100,"categories":101,"keywords":102,"publishedAt":19},"01M79G1H6WN0Q6C0VCAXCZX6B7","vss-query-analytics","Use this skill when reading video-analytics metrics, incidents, alerts, and sensor data via the VA-MCP server (port 9901). Not for live VLM or incident-range narrative reports.","https:\u002F\u002Fgithub.com\u002FElogic360\u002Fsoftware-factory\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Fvss-query-analytics\u002FSKILL.md",[],[],{"skillNo":104,"name":105,"nameZh":6,"description":106,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":107,"categories":108,"keywords":109,"publishedAt":19},"01M79G1DYPNXQS7QHZ0AAF5FX5","vss-manage-video-io-storage","Use to call the VIOS REST API (sensor list, timelines, clip extraction, snapshots, add\u002Fdelete sensors and streams). Not for VLM inference or search.","https:\u002F\u002Fgithub.com\u002FElogic360\u002Fsoftware-factory\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Fvss-manage-video-io-storage\u002FSKILL.md",[],[]]