[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7977NB0DN0TFZM6NJFVG0ME":3,"skill-same:01M7977NB0DN0TFZM6NJFVG0ME":71},{"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":27,"commitSha":28,"skillDirectory":29,"forksCount":30,"repositoryHtmlUrl":31,"repositoryHomepageUrl":6,"ownerHtmlUrl":32,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":33,"install":34,"resources":46},"01M7977NB0DN0TFZM6NJFVG0ME","vllm-omni",null,"Use vLLM-Omni for omni-modality model inference and serving: local Omni APIs, OpenAI-compatible --omni servers, stage deploy configs, model recipes, and model integration workflows.","VectorSpaceLab\u002FAREX-Skill","VectorSpaceLab","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F180637102?v=4","01M7538C5WPSJGWF7KHR4EFZ9F","01M7538C6R2GFP53NPXTCZNBRT",351,"Apache-2.0","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fvllm-omni\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: vllm-omni\ndescription: \"Use vLLM-Omni for omni-modality model inference and serving: local\n  Omni APIs, OpenAI-compatible --omni servers, stage deploy configs, model\n  recipes, and model integration workflows.\"\ndisable-model-invocation: true\nmetadata:\n  disco-role: operating\nlicense: Apache 2.0\n---\n\n# vLLM-Omni\n\nUse this repo skill when a task involves vLLM-Omni: omni-modality inference or serving for text, image, audio, video, action, diffusion, TTS, OpenAI-compatible APIs, stage-based deployment, supported model recipes, or source-level model integration.\n\n## Quick start route\n\n1. For install\u002Fimport\u002Fbackend checks, read [install-and-backends.md](references\u002Finstall-and-backends.md) and run the safe checker when needed:\n\n   ```bash\n   python scripts\u002Fcheck_environment.py --require-vllm 0.26\n   ```\n\n   Add `--require-cuda` only when live GPU generation or serving must be verified.\n2. Choose the closest sub-skill:\n\n   | Task | Read |\n   | --- | --- |\n   | Local Python scripts using `Omni`, `AsyncOmni`, prompt dictionaries, sampling params, or multimodal outputs | [offline-inference](sub-skills\u002Foffline-inference\u002FSKILL.md) |\n   | `vllm serve ... --omni`, OpenAI-compatible HTTP payloads, curl\u002FOpenAI SDK clients, realtime\u002Fstreaming, stage head\u002Fheadless launch | [online-serving](sub-skills\u002Fonline-serving\u002FSKILL.md) |\n   | Deploy YAML overlays, `stages`, connectors, memory placement, `--stage-overrides`, multi-node\u002Fstage planning | [stage-configuration](sub-skills\u002Fstage-configuration\u002FSKILL.md) |\n   | Choosing supported model families, endpoints, hardware backends, diffusion controls, quantization\u002Foffload\u002Fcache, benchmark metrics | [model-recipes](sub-skills\u002Fmodel-recipes\u002FSKILL.md) |\n   | Adding custom pipelines, model registrations, TTS adapters, or selecting focused maintainer tests | [model-integration](sub-skills\u002Fmodel-integration\u002FSKILL.md) |\n3. If a failure crosses workflows, read [troubleshooting.md](references\u002Ftroubleshooting.md) before running full model examples.\n4. To check whether this skill matches the source checkout, read [repo-provenance.md](references\u002Frepo-provenance.md).\n\n## What vLLM-Omni adds\n\nvLLM-Omni extends vLLM's text-oriented autoregressive runtime with stage-based and diffusion\u002Fgeneration paths for omni-modality workloads:\n\n- multimodal chat and reasoning with text, image, audio, video, and action inputs;\n- heterogeneous outputs including text tokens, images, audio, video, latents, trajectories, and custom payloads;\n- OpenAI-compatible serving with Omni-specific request fields and endpoints;\n- stage-based deploy configs, connectors, head\u002Fheadless worker launches, and async chunking;\n- diffusion image\u002Fvideo\u002Faudio pipelines with attention, parallelism, offload, cache, quantization, and LoRA controls;\n- TTS and realtime\u002Fduplex audio routes;\n- model integration surfaces for custom pipelines and supported model families.\n\n## Install\u002Fbackend guardrails\n\n- Use a fresh environment. vLLM, PyTorch, CUDA\u002FROCm\u002Fvendor libraries, media packages, and attention kernels are easy to mix incorrectly.\n- Align upstream `vllm` with the vLLM-Omni release line. For this checkout's docs, `vllm==0.26.x` is the expected upstream line.\n- Set `VLLM_OMNI_TARGET_DEVICE` during source installs when automation must choose CUDA, ROCm, NPU, XPU, MUSA, or CPU requirements deterministically.\n- CPU parser\u002Fconfig checks do not prove live model serving. Full generation usually requires GPU\u002Faccelerator hardware, model weights, license\u002Fcache access, and runtime budget.\n- Do not run broad native examples or benchmarks by default; use bundled no-network helpers first.\n\n## Common entry points\n\n- Package import: `import vllm_omni` for registration side effects and `from vllm_omni.entrypoints.omni import Omni` for local generation.\n- Local API: `Omni(model=..., **kwargs).generate(...)` and `AsyncOmni(model=...).generate(...)`.\n- CLI serving: `vllm serve MODEL --omni ...` or `vllm-omni serve MODEL --omni ...`.\n- Deploy config concepts: `DeployConfig`, `StageDeployConfig`, `PipelineConfig`, `stage_id`, `devices`, connectors, `gpu_memory_utilization`, `async_chunk`, platform overrides.\n- Model\u002Frecipe concepts: Qwen3-Omni, Qwen2.5-Omni, Qwen-Image\u002FZ-Image\u002FGLM\u002FHunyuan\u002FWan\u002FLTX\u002FCosmos3\u002FMiniMax-H3, Qwen3-TTS\u002FVoxCPM2\u002FMOSS\u002FHiggs\u002FIndexTTS, DreamZero\u002FGR00T\u002FInternVLA.\n\n## Bundled helpers\n\n| Helper | Use |\n| --- | --- |\n| `scripts\u002Fcheck_environment.py` | Verify package metadata, imports, optional CUDA, and Omni CLI help without model downloads. |\n| `sub-skills\u002Foffline-inference\u002Fscripts\u002Fbuild_offline_request.py` | Generate safe local Python request snippets without importing vLLM-Omni or loading a model. |\n| `sub-skills\u002Fonline-serving\u002Fscripts\u002Fbuild_openai_payload.py` | Generate curl\u002FOpenAI SDK payload scaffolds without sending HTTP requests. |\n| `sub-skills\u002Fstage-configuration\u002Fscripts\u002Fvalidate_deploy_yaml.py` | Sanity-check deploy YAML shape and connector references without model loading. |\n| `sub-skills\u002Fstage-configuration\u002Fscripts\u002Fplan_stage_memory.py` | Estimate stage placement and memory utilization from user-provided GPU\u002Fstage counts. |\n| `sub-skills\u002Fmodel-recipes\u002Fscripts\u002Fquery_model_catalog.py` | Query the bundled representative model catalog by task\u002Fbackend. |\n| `sub-skills\u002Fmodel-integration\u002Fscripts\u002Fcheck_tts_adapter_contract.py` | Statically review TTS adapter source shape without imports or model execution. |\n\n## When to refresh this skill\n\nRefresh when vLLM-Omni changes public CLI flags, OpenAI protocol fields, deploy schemas, supported model tables, diffusion controls, TTS adapters, model registry behavior, or aligned upstream vLLM version.\n","# vLLM-Omni\n\nUse this repo skill when a task involves vLLM-Omni: omni-modality inference or serving for text, image, audio, video, action, diffusion, TTS, OpenAI-compatible APIs, stage-based deployment, supported model recipes, or source-level model integration.\n\n## Quick start route\n\n1. For install\u002Fimport\u002Fbackend checks, read [install-and-backends.md](references\u002Finstall-and-backends.md) and run the safe checker when needed:\n\n   ```bash\n   python scripts\u002Fcheck_environment.py --require-vllm 0.26\n   ```\n\n   Add `--require-cuda` only when live GPU generation or serving must be verified.\n2. Choose the closest sub-skill:\n\n   | Task | Read |\n   | --- | --- |\n   | Local Python scripts using `Omni`, `AsyncOmni`, prompt dictionaries, sampling params, or multimodal outputs | [offline-inference](sub-skills\u002Foffline-inference\u002FSKILL.md) |\n   | `vllm serve ... --omni`, OpenAI-compatible HTTP payloads, curl\u002FOpenAI SDK clients, realtime\u002Fstreaming, stage head\u002Fheadless launch | [online-serving](sub-skills\u002Fonline-serving\u002FSKILL.md) |\n   | Deploy YAML overlays, `stages`, connectors, memory placement, `--stage-overrides`, multi-node\u002Fstage planning | [stage-configuration](sub-skills\u002Fstage-configuration\u002FSKILL.md) |\n   | Choosing supported model families, endpoints, hardware backends, diffusion controls, quantization\u002Foffload\u002Fcache, benchmark metrics | [model-recipes](sub-skills\u002Fmodel-recipes\u002FSKILL.md) |\n   | Adding custom pipelines, model registrations, TTS adapters, or selecting focused maintainer tests | [model-integration](sub-skills\u002Fmodel-integration\u002FSKILL.md) |\n3. If a failure crosses workflows, read [troubleshooting.md](references\u002Ftroubleshooting.md) before running full model examples.\n4. To check whether this skill matches the source checkout, read [repo-provenance.md](references\u002Frepo-provenance.md).\n\n## What vLLM-Omni adds\n\nvLLM-Omni extends vLLM's text-oriented autoregressive runtime with stage-based and diffusion\u002Fgeneration paths for omni-modality workloads:\n\n- multimodal chat and reasoning with text, image, audio, video, and action inputs;\n- heterogeneous outputs including text tokens, images, audio, video, latents, trajectories, and custom payloads;\n- OpenAI-compatible serving with Omni-specific request fields and endpoints;\n- stage-based deploy configs, connectors, head\u002Fheadless worker launches, and async chunking;\n- diffusion image\u002Fvideo\u002Faudio pipelines with attention, parallelism, offload, cache, quantization, and LoRA controls;\n- TTS and realtime\u002Fduplex audio routes;\n- model integration surfaces for custom pipelines and supported model families.\n\n## Install\u002Fbackend guardrails\n\n- Use a fresh environment. vLLM, PyTorch, CUDA\u002FROCm\u002Fvendor libraries, media packages, and attention kernels are easy to mix incorrectly.\n- Align upstream `vllm` with the vLLM-Omni release line. For this checkout's docs, `vllm==0.26.x` is the expected upstream line.\n- Set `VLLM_OMNI_TARGET_DEVICE` during source installs when automation must choose CUDA, ROCm, NPU, XPU, MUSA, or CPU requirements deterministically.\n- CPU parser\u002Fconfig checks do not prove live model serving. Full generation usually requires GPU\u002Faccelerator hardware, model weights, license\u002Fcache access, and runtime budget.\n- Do not run broad native examples or benchmarks by default; use bundled no-network helpers first.\n\n## Common entry points\n\n- Package import: `import vllm_omni` for registration side effects and `from vllm_omni.entrypoints.omni import Omni` for local generation.\n- Local API: `Omni(model=..., **kwargs).generate(...)` and `AsyncOmni(model=...).generate(...)`.\n- CLI serving: `vllm serve MODEL --omni ...` or `vllm-omni serve MODEL --omni ...`.\n- Deploy config concepts: `DeployConfig`, `StageDeployConfig`, `PipelineConfig`, `stage_id`, `devices`, connectors, `gpu_memory_utilization`, `async_chunk`, platform overrides.\n- Model\u002Frecipe concepts: Qwen3-Omni, Qwen2.5-Omni, Qwen-Image\u002FZ-Image\u002FGLM\u002FHunyuan\u002FWan\u002FLTX\u002FCosmos3\u002FMiniMax-H3, Qwen3-TTS\u002FVoxCPM2\u002FMOSS\u002FHiggs\u002FIndexTTS, DreamZero\u002FGR00T\u002FInternVLA.\n\n## Bundled helpers\n\n| Helper | Use |\n| --- | --- |\n| `scripts\u002Fcheck_environment.py` | Verify package metadata, imports, optional CUDA, and Omni CLI help without model downloads. |\n| `sub-skills\u002Foffline-inference\u002Fscripts\u002Fbuild_offline_request.py` | Generate safe local Python request snippets without importing vLLM-Omni or loading a model. |\n| `sub-skills\u002Fonline-serving\u002Fscripts\u002Fbuild_openai_payload.py` | Generate curl\u002FOpenAI SDK payload scaffolds without sending HTTP requests. |\n| `sub-skills\u002Fstage-configuration\u002Fscripts\u002Fvalidate_deploy_yaml.py` | Sanity-check deploy YAML shape and connector references without model loading. |\n| `sub-skills\u002Fstage-configuration\u002Fscripts\u002Fplan_stage_memory.py` | Estimate stage placement and memory utilization from user-provided GPU\u002Fstage counts. |\n| `sub-skills\u002Fmodel-recipes\u002Fscripts\u002Fquery_model_catalog.py` | Query the bundled representative model catalog by task\u002Fbackend. |\n| `sub-skills\u002Fmodel-integration\u002Fscripts\u002Fcheck_tts_adapter_contract.py` | Statically review TTS adapter source shape without imports or model execution. |\n\n## When to refresh this skill\n\nRefresh when vLLM-Omni changes public CLI flags, OpenAI protocol fields, deploy schemas, supported model tables, diffusion controls, TTS adapters, model registry behavior, or aligned upstream vLLM version.\n",{"name":5,"license":23,"metadata":24,"description":7,"disable-model-invocation":26},"Apache 2.0",{"disco-role":25},"operating",true,"main","ac3fe1afa80fb9a09775ecfb2b6cc3ba850a2db6","skills\u002Frepositories\u002Frepo-skills\u002Fvllm-omni",27,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab","2026-09-03 10:25:21",{"source":8,"skillName":5,"commands":35,"prompt":45,"zipAvailable":26},[36,39,42],{"tool":37,"command":38},"npx","npx skills add VectorSpaceLab\u002FAREX-Skill --skill vllm-omni",{"tool":40,"command":41},"bunx","bunx skills add VectorSpaceLab\u002FAREX-Skill --skill vllm-omni",{"tool":43,"command":44},"pnpm","pnpm dlx skills add VectorSpaceLab\u002FAREX-Skill --skill vllm-omni","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fvllm-omni\u002FSKILL.md",[47,53,57,61,65],{"path":48,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":51,"sourceHtmlUrl":52},"references\u002Finstall-and-backends.md","REFERENCE","参考文档",4942,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fvllm-omni\u002Freferences\u002Finstall-and-backends.md",{"path":54,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":55,"sourceHtmlUrl":56},"references\u002Frepo-provenance.md",2377,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fvllm-omni\u002Freferences\u002Frepo-provenance.md",{"path":58,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":59,"sourceHtmlUrl":60},"references\u002Frepo-routing-metadata.json",505,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fvllm-omni\u002Freferences\u002Frepo-routing-metadata.json",{"path":62,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":63,"sourceHtmlUrl":64},"references\u002Ftroubleshooting.md",4237,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fvllm-omni\u002Freferences\u002Ftroubleshooting.md",{"path":66,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":69,"sourceHtmlUrl":70},"scripts\u002Fcheck_environment.py","SCRIPT","脚本",6709,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fvllm-omni\u002Fscripts\u002Fcheck_environment.py",[72,79,86,93,100,107],{"skillNo":73,"name":74,"nameZh":6,"description":75,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":76,"categories":77,"keywords":78,"publishedAt":18},"01M799NYSH1R3RVAHSM2D88Y84","passnet-triton-opt","Make a single PassNet Triton kernel fast AND numerically faithful: performance model (when a replacement can win at all), block\u002Fgrid\u002Fwarp tuning, autotune policy, launch overhead, and per-op numeric recipes to pass the dtype baseline tolerances. Use when a pass matches and is correct but speedup below expected, or when correctness fails by small numeric margins.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-triton-opt\u002FSKILL.md",[],[],{"skillNo":80,"name":81,"nameZh":6,"description":82,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":83,"categories":84,"keywords":85,"publishedAt":18},"01M799NS5APEC0N1FZ7XZVW6BN","passnet-solve","END-TO-END playbook for solving one PassNet sample: analyze the computation graphs, decide the optimization strategy (what to fuse, what to replace, what to leave alone), drive the iteration loop, and maximize the sample score. This is the ENTRY skill — invoke it first for any PassNet optimization task; it tells you when to use passnet-pattern-fusion, passnet-triton-opt and passnet-feedback.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-solve\u002FSKILL.md",[],[],{"skillNo":87,"name":88,"nameZh":6,"description":89,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":90,"categories":91,"keywords":92,"publishedAt":18},"01M799NNY9BKSM21V5R7QVTBJX","passnet-skill","PassNet GPU kernel optimization via compiler passes. Design and implement Triton-based optimization passes, create pass files under .\u002Fpass_dir\u002F, self-evaluate with pass_evaluator, and iterate to maximize GPU speedup.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-skill\u002FSKILL.md",[],[],{"skillNo":94,"name":95,"nameZh":6,"description":96,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":97,"categories":98,"keywords":99,"publishedAt":18},"01M799NHH6NP7Z1RK77KGXAN6S","passnet-pattern-fusion","Author PassNet pass files: write patterns that actually MATCH the FX graph, pick fusion regions (which ops to absorb into one Triton kernel), and structure multi-pass submissions with the mandatory shared-dispatch architecture. Use when creating or fixing pattern()\u002Freplacement_args()\u002Freplacement_func() files, when a pass \"failed to match\", or when deciding how to fuse multiple kernels\u002Fops.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-pattern-fusion\u002FSKILL.md",[],[],{"skillNo":101,"name":102,"nameZh":6,"description":103,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":104,"categories":105,"keywords":106,"publishedAt":18},"01M799ND2J4WNC2PGAZ0A06NDF","passnet-orchestrate","Round\u002Fbatch planning and decision support for PassNet graph optimization. Use for multi-sample triage, worker allocation, eval-budget policy, and post-eval keep\u002Frevert decisions. For solving one concrete sample end to end, use passnet-solve as the entry skill; consult this skill only when you need broader planning or a specific decision gate.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-orchestrate\u002FSKILL.md",[],[],{"skillNo":108,"name":109,"nameZh":6,"description":110,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":111,"categories":112,"keywords":113,"publishedAt":18},"01M799N8SBTMY02ZDG2AZF5ZWE","passnet-feedback","Fast, deterministic feedback for PassNet work: pre-flight pattern\u002Fmatch verification WITHOUT burning a GPU evaluation, per-node bottleneck analysis of a sample's graphs, and eval-log parsing into per-variant status + estimated score + failure classification. Use BEFORE every GPU evaluation (check_pattern), at the START of a sample (analyze_graph), and AFTER every evaluation (parse_eval_log).","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-feedback\u002FSKILL.md",[],[]]