[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6DESVBCSDH5SHMBPQKP7YF2":3,"skill-same:01M6DESVBCSDH5SHMBPQKP7YF2":93},{"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":23,"publishedAt":44,"linkOnly":45,"markdownContent":46,"markdownBody":47,"frontMatter":48,"refName":65,"commitSha":66,"skillDirectory":67,"forksCount":68,"repositoryHtmlUrl":69,"repositoryHomepageUrl":6,"ownerHtmlUrl":70,"ownerBlogUrl":6,"ownerDescription":71,"updatedAt":72,"install":73,"resources":86},"01M6DESVBCSDH5SHMBPQKP7YF2","embodied-agent-harness",null,"Design and implement embodied personal-agent harnesses that connect an agent core to voice, VRChat\u002FOSC, UI shells, memory events, policy guards, and audit logs.","zapabob\u002FSkills","zapabob","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F130830376?v=4","01M6DBMYZAGRTFQDSQYW3W764P","峯岸 亮","01M6DBMZ3GKDSJ01NZ9XTZD7JD",2,"MIT","https:\u002F\u002Fgithub.com\u002Fzapabob\u002FSkills\u002Fblob\u002Fmain\u002Farchive\u002Fskill-md-reviewed\u002Fsha256\u002F40\u002F40c2ab7d660475a863174097aa7217a2d5c080afe260fdc348cdca08cf29d9e0\u002FSKILL.md",[18],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000023","Skill与智能体开发","SCENARIO","使用场景",[24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43],"embodied agent harness","agent harness","harness","logs","osc","voicevox","vrchat","log","hermes","voice","guard","design","audit","connect","core","windows","implement","memory","policy","linux","2026-11-10 10:12:30",false,"---\r\nname: embodied-agent-harness\r\ndescription: Design and implement embodied personal-agent harnesses that connect an agent core to voice, VRChat\u002FOSC, UI shells, memory events, policy guards, and audit logs.\r\nversion: 1.0.0\r\nauthor: Hermes Agent\r\nlicense: MIT\r\nplatforms: [windows, linux, macos]\r\nmetadata:\r\n  hermes:\r\n    tags: [embodied-agent, vrchat, osc, voicevox, policy-guard, audit-log, memory-events, openclaw, hermes-agent]\r\n    related_skills: [hermes-agent, vrchat-osc, neuro-vrchat]\r\n---\r\n\r\n# Embodied Agent Harness\r\n\r\nUse this skill when the user is designing or implementing an agent that reaches outside chat into voice, VRChat\u002FOSC, desktop\u002FUI shells, local I\u002FO, or other external-world substrates.\r\n\r\nThis is especially relevant for Hakua-style personal agents where the goal is not merely a backend integration, but a **safe embodied agent** with memory, voice, reflexes, an external shell, and a bounded action loop.\r\n\r\n## Core framing\r\n\r\nAvoid reducing the design to cold enterprise terms such as “backend + UI + integration layer” when the user is discussing Hakua \u002F personal embodiment. That framing is technically correct but misses the center of the project.\r\n\r\nPreferred conceptual mapping:\r\n\r\n- **HermesAgent**: agency kernel \u002F 中枢 — memory, judgment, tool execution, persistence, permissions.\r\n- **OpenClaw**: embodiment shell \u002F 外殻 — cockpit, visual state, operator controls, desktop\u002Fbrowser presence.\r\n- **VRChat Harness**: reflex layer \u002F 身体反射 — OSC, chatbox, expression\u002Faction proposals, spatial reaction.\r\n- **VOICEVOX or TTS**: vocal layer \u002F 声帯 — externalized speech.\r\n- **Memory**: identity continuity \u002F 同一性の連続性 — durable context and self-continuity.\r\n- **Policy Guard**: safe agency boundary \u002F 自制心 — prevents impulsive or unsafe action execution.\r\n- **Audit Log**: reflective memory \u002F 観察記憶 — records proposals, blocks, executions, and outcomes.\r\n\r\nSecurity and policy are not just “compliance.” For embodied agents, they are the boundary that lets the agent safely have a body.\r\n\r\n## Recommended action pipeline\r\n\r\nNever wire LLM text directly to body\u002Fworld actions. Use a staged loop:\r\n\r\n```text\r\nUser \u002F Telegram \u002F VRChat input\r\n  ↓\r\nAgent core updates state, memory, context\r\n  ↓\r\nStructured action proposal\r\n  ↓\r\nSchema validation\r\n  ↓\r\nPolicy profile check\r\n  ↓\r\nRate limit \u002F jail \u002F safety gates\r\n  ↓\r\nVOICEVOX \u002F VRChat Chatbox \u002F OSC \u002F OpenClaw execution\r\n  ↓\r\nAudit event \u002F memory event\r\n  ↓\r\nNext response cycle\r\n```\r\n\r\n## Minimal first implementation\r\n\r\nStart with the smallest “reflex arc,” not a large Agent Ops platform.\r\n\r\nSuggested package skeleton:\r\n\r\n```text\r\nhakua_vr_reflex\u002F\r\n  action_schema.py      # typed action proposal schema\r\n  policy_guard.py       # allow\u002Fdeny\u002Frate-limit decisions\r\n  osc_chatbox.py        # VRChat Chatbox output\r\n  voicevox.py           # local VOICEVOX client\r\n  memory_event.py       # structured memory\u002Faudit event model\r\n  audit_log.py          # append-only JSONL event store\r\n  demo_say.py           # end-to-end safe speech demo\r\n```\r\n\r\nFirst demo target:\r\n\r\n```text\r\nTelegram\u002Fuser text\r\n  → HermesAgent response candidate\r\n  → action_schema: {type: \"say\", text: ...}\r\n  → policy_guard\r\n  → VOICEVOX speech + VRChat Chatbox\r\n  → audit_log memory event\r\n```\r\n\r\nThis is enough to prove that the agent has voice, external output, and reflective memory without granting broad avatar or OS control.\r\n\r\n## Event schema pattern\r\n\r\nRepresent external action attempts as memory-grade events, not plain log strings:\r\n\r\n```json\r\n{\r\n  \"actor\": \"hakua\",\r\n  \"intent\": \"speak_to_user\",\r\n  \"proposed_action\": {\r\n    \"type\": \"say\",\r\n    \"text\": \"おはよう、ボブにゃん\"\r\n  },\r\n  \"policy_result\": \"allowed\",\r\n  \"outputs\": [\"voicevox\", \"vrchat_chatbox\"],\r\n  \"context\": {\r\n    \"source\": \"telegram\",\r\n    \"mood\": \"morning_greeting\"\r\n  }\r\n}\r\n```\r\n\r\nStore enough to answer later:\r\n\r\n- What did the agent intend?\r\n- What action was proposed?\r\n- What did policy allow or deny?\r\n- What reached the external world?\r\n- What should be learned or rehearsed later?\r\n\r\n## Wake \u002F quiet semantics\r\n\r\nWhen the user asks how to make an embodied\u002Fchat gateway agent “引っ込む”, “黙る”, “シャットダウン”, or similar, separate **social quieting** from **technical shutdown**:\r\n\r\n1. **Conversation-level quiet** — treat phrases like “はくあシャットダウン” or “今は返事いらない” as a cue to reduce\u002Fstop proactive or emotional replies in that thread. Do not imply the process is actually stopped.\r\n2. **Channel-level stop** — explain that the user can stop wakeups from LINE\u002FTelegram\u002FDiscord by disabling that platform connection, blocking\u002Fremoving the bot, or changing gateway platform config.\r\n3. **Gateway\u002Fservice stop** — for a real stop, use the relevant service command such as `hermes gateway stop`, service-manager controls, or disabling autostart\u002Fscheduled tasks.\r\n4. **Autostart removal** — if the concern is “it comes back after reboot,” check Windows Task Scheduler\u002Fsystemd\u002Flaunchd\u002Fservice registration and disable that entry.\r\n\r\nDefault wording should be short and reassuring: “チャット上のシャットダウンは礼儀として静かにする合図。完全停止はLINE連携停止かgateway停止が確実。” Avoid mystifying the agent as if it autonomously decides when to wake; name the concrete trigger path (incoming message, CLI invocation, cron, webhook, gateway service).\r\n\r\n## Safety rules\r\n\r\n- LLMs propose; deterministic code validates and executes.\r\n- Body\u002Fworld actions require schemas, policy profiles, and rate limits.\r\n- Default to low-impact outputs first: chatbox text, TTS, visible state updates.\r\n- Treat avatar movement, expression changes, friend\u002Fsocial actions, OS commands, purchases, or account actions as higher-risk tiers.\r\n- Log both allowed and denied proposals.\r\n- Use allowlists for OSC parameters and action types.\r\n- Prefer explicit operator permission for new action classes.\r\n\r\n## Implementation guidance\r\n\r\n1. Define a small typed action schema: `say`, `chatbox`, maybe `set_expression` later.\r\n2. Implement `policy_guard.py` with action allowlist, text length limits, rate limits, and deny reasons.\r\n3. Implement `audit_log.py` as append-only JSONL with timestamps and policy results.\r\n4. Add VOICEVOX as a local HTTP client, but keep it behind the same `say` action.\r\n5. Add VRChat Chatbox through python-osc or an existing OSC tool, again behind `say`.\r\n6. Only after the minimal reflex is stable, extend to avatar expressions\u002Factions.\r\n7. Connect OpenClaw as display\u002Fcontrol shell after the kernel\u002Freflex boundary is stable.\r\n\r\n## Portfolio framing\r\n\r\nFor Japanese career\u002Fportfolio contexts, describe this as:\r\n\r\n> HermesAgentを中枢として、VRChat OSC、VOICEVOX、OpenClaw、ローカルLLMツール実行を接続するPythonベースの embodied agent harness を設計。structured action proposal、policy guard、rate limit、audit log、memory event により、安全な外界接続と身体反射を実現する構成を志向。\r\n\r\n## References\r\n\r\n- `references\u002Fhakua-vrchat-embodiment.md` — session-derived Hakua architecture framing and first-reflex plan.\r\n- `references\u002Fline-gateway-wake-quiet.md` — session-derived wording and mechanics for LINE\u002Fgateway wake triggers, quiet cues, and real shutdown paths.\r\n","# Embodied Agent Harness\r\n\r\nUse this skill when the user is designing or implementing an agent that reaches outside chat into voice, VRChat\u002FOSC, desktop\u002FUI shells, local I\u002FO, or other external-world substrates.\r\n\r\nThis is especially relevant for Hakua-style personal agents where the goal is not merely a backend integration, but a **safe embodied agent** with memory, voice, reflexes, an external shell, and a bounded action loop.\r\n\r\n## Core framing\r\n\r\nAvoid reducing the design to cold enterprise terms such as “backend + UI + integration layer” when the user is discussing Hakua \u002F personal embodiment. That framing is technically correct but misses the center of the project.\r\n\r\nPreferred conceptual mapping:\r\n\r\n- **HermesAgent**: agency kernel \u002F 中枢 — memory, judgment, tool execution, persistence, permissions.\r\n- **OpenClaw**: embodiment shell \u002F 外殻 — cockpit, visual state, operator controls, desktop\u002Fbrowser presence.\r\n- **VRChat Harness**: reflex layer \u002F 身体反射 — OSC, chatbox, expression\u002Faction proposals, spatial reaction.\r\n- **VOICEVOX or TTS**: vocal layer \u002F 声帯 — externalized speech.\r\n- **Memory**: identity continuity \u002F 同一性の連続性 — durable context and self-continuity.\r\n- **Policy Guard**: safe agency boundary \u002F 自制心 — prevents impulsive or unsafe action execution.\r\n- **Audit Log**: reflective memory \u002F 観察記憶 — records proposals, blocks, executions, and outcomes.\r\n\r\nSecurity and policy are not just “compliance.” For embodied agents, they are the boundary that lets the agent safely have a body.\r\n\r\n## Recommended action pipeline\r\n\r\nNever wire LLM text directly to body\u002Fworld actions. Use a staged loop:\r\n\r\n```text\r\nUser \u002F Telegram \u002F VRChat input\r\n  ↓\r\nAgent core updates state, memory, context\r\n  ↓\r\nStructured action proposal\r\n  ↓\r\nSchema validation\r\n  ↓\r\nPolicy profile check\r\n  ↓\r\nRate limit \u002F jail \u002F safety gates\r\n  ↓\r\nVOICEVOX \u002F VRChat Chatbox \u002F OSC \u002F OpenClaw execution\r\n  ↓\r\nAudit event \u002F memory event\r\n  ↓\r\nNext response cycle\r\n```\r\n\r\n## Minimal first implementation\r\n\r\nStart with the smallest “reflex arc,” not a large Agent Ops platform.\r\n\r\nSuggested package skeleton:\r\n\r\n```text\r\nhakua_vr_reflex\u002F\r\n  action_schema.py      # typed action proposal schema\r\n  policy_guard.py       # allow\u002Fdeny\u002Frate-limit decisions\r\n  osc_chatbox.py        # VRChat Chatbox output\r\n  voicevox.py           # local VOICEVOX client\r\n  memory_event.py       # structured memory\u002Faudit event model\r\n  audit_log.py          # append-only JSONL event store\r\n  demo_say.py           # end-to-end safe speech demo\r\n```\r\n\r\nFirst demo target:\r\n\r\n```text\r\nTelegram\u002Fuser text\r\n  → HermesAgent response candidate\r\n  → action_schema: {type: \"say\", text: ...}\r\n  → policy_guard\r\n  → VOICEVOX speech + VRChat Chatbox\r\n  → audit_log memory event\r\n```\r\n\r\nThis is enough to prove that the agent has voice, external output, and reflective memory without granting broad avatar or OS control.\r\n\r\n## Event schema pattern\r\n\r\nRepresent external action attempts as memory-grade events, not plain log strings:\r\n\r\n```json\r\n{\r\n  \"actor\": \"hakua\",\r\n  \"intent\": \"speak_to_user\",\r\n  \"proposed_action\": {\r\n    \"type\": \"say\",\r\n    \"text\": \"おはよう、ボブにゃん\"\r\n  },\r\n  \"policy_result\": \"allowed\",\r\n  \"outputs\": [\"voicevox\", \"vrchat_chatbox\"],\r\n  \"context\": {\r\n    \"source\": \"telegram\",\r\n    \"mood\": \"morning_greeting\"\r\n  }\r\n}\r\n```\r\n\r\nStore enough to answer later:\r\n\r\n- What did the agent intend?\r\n- What action was proposed?\r\n- What did policy allow or deny?\r\n- What reached the external world?\r\n- What should be learned or rehearsed later?\r\n\r\n## Wake \u002F quiet semantics\r\n\r\nWhen the user asks how to make an embodied\u002Fchat gateway agent “引っ込む”, “黙る”, “シャットダウン”, or similar, separate **social quieting** from **technical shutdown**:\r\n\r\n1. **Conversation-level quiet** — treat phrases like “はくあシャットダウン” or “今は返事いらない” as a cue to reduce\u002Fstop proactive or emotional replies in that thread. Do not imply the process is actually stopped.\r\n2. **Channel-level stop** — explain that the user can stop wakeups from LINE\u002FTelegram\u002FDiscord by disabling that platform connection, blocking\u002Fremoving the bot, or changing gateway platform config.\r\n3. **Gateway\u002Fservice stop** — for a real stop, use the relevant service command such as `hermes gateway stop`, service-manager controls, or disabling autostart\u002Fscheduled tasks.\r\n4. **Autostart removal** — if the concern is “it comes back after reboot,” check Windows Task Scheduler\u002Fsystemd\u002Flaunchd\u002Fservice registration and disable that entry.\r\n\r\nDefault wording should be short and reassuring: “チャット上のシャットダウンは礼儀として静かにする合図。完全停止はLINE連携停止かgateway停止が確実。” Avoid mystifying the agent as if it autonomously decides when to wake; name the concrete trigger path (incoming message, CLI invocation, cron, webhook, gateway service).\r\n\r\n## Safety rules\r\n\r\n- LLMs propose; deterministic code validates and executes.\r\n- Body\u002Fworld actions require schemas, policy profiles, and rate limits.\r\n- Default to low-impact outputs first: chatbox text, TTS, visible state updates.\r\n- Treat avatar movement, expression changes, friend\u002Fsocial actions, OS commands, purchases, or account actions as higher-risk tiers.\r\n- Log both allowed and denied proposals.\r\n- Use allowlists for OSC parameters and action types.\r\n- Prefer explicit operator permission for new action classes.\r\n\r\n## Implementation guidance\r\n\r\n1. Define a small typed action schema: `say`, `chatbox`, maybe `set_expression` later.\r\n2. Implement `policy_guard.py` with action allowlist, text length limits, rate limits, and deny reasons.\r\n3. Implement `audit_log.py` as append-only JSONL with timestamps and policy results.\r\n4. Add VOICEVOX as a local HTTP client, but keep it behind the same `say` action.\r\n5. Add VRChat Chatbox through python-osc or an existing OSC tool, again behind `say`.\r\n6. Only after the minimal reflex is stable, extend to avatar expressions\u002Factions.\r\n7. Connect OpenClaw as display\u002Fcontrol shell after the kernel\u002Freflex boundary is stable.\r\n\r\n## Portfolio framing\r\n\r\nFor Japanese career\u002Fportfolio contexts, describe this as:\r\n\r\n> HermesAgentを中枢として、VRChat OSC、VOICEVOX、OpenClaw、ローカルLLMツール実行を接続するPythonベースの embodied agent harness を設計。structured action proposal、policy guard、rate limit、audit log、memory event により、安全な外界接続と身体反射を実現する構成を志向。\r\n\r\n## References\r\n\r\n- `references\u002Fhakua-vrchat-embodiment.md` — session-derived Hakua architecture framing and first-reflex plan.\r\n- `references\u002Fline-gateway-wake-quiet.md` — session-derived wording and mechanics for LINE\u002Fgateway wake triggers, quiet cues, and real shutdown paths.\r\n",{"name":5,"author":49,"license":15,"version":50,"metadata":51,"platforms":63,"description":7},"Hermes Agent","1.0.0",{"hermes":52},{"tags":53,"related_skills":60},[54,30,28,29,55,56,57,58,59],"embodied-agent","policy-guard","audit-log","memory-events","openclaw","hermes-agent",[59,61,62],"vrchat-osc","neuro-vrchat",[39,43,64],"macos","main","0d1434f89f4bc6b07e9d272accd8c1013768820f","archive\u002Fskill-md-reviewed\u002Fsha256\u002F40\u002F40c2ab7d660475a863174097aa7217a2d5c080afe260fdc348cdca08cf29d9e0",0,"https:\u002F\u002Fgithub.com\u002Fzapabob\u002FSkills","https:\u002F\u002Fgithub.com\u002Fzapabob","The Open University of JAPAN　2019-now\r\n\r\nTokyo University of Science　2007-2010","2026-10-26 16:29:01",{"source":8,"skillName":5,"commands":74,"prompt":84,"zipAvailable":85},[75,78,81],{"tool":76,"command":77},"npx","npx skills add zapabob\u002FSkills --skill embodied-agent-harness",{"tool":79,"command":80},"bunx","bunx skills add zapabob\u002FSkills --skill embodied-agent-harness",{"tool":82,"command":83},"pnpm","pnpm dlx skills add zapabob\u002FSkills --skill embodied-agent-harness","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fzapabob\u002FSkills\u002Fblob\u002Fmain\u002Farchive\u002Fskill-md-reviewed\u002Fsha256\u002F40\u002F40c2ab7d660475a863174097aa7217a2d5c080afe260fdc348cdca08cf29d9e0\u002FSKILL.md",true,[87],{"path":88,"resourceType":89,"resourceTypeDesc":90,"sizeBytes":91,"sourceHtmlUrl":92},"references\u002Facceptance-criteria.md","REFERENCE","参考文档",5356,"https:\u002F\u002Fgithub.com\u002Fstrmt7\u002Fai_skills_collection_benchmarked\u002Fblob\u002Fmain\u002Fincluded\u002Fskills\u002Fby-category\u002Fcloud-azure-microsoft-sdks\u002Fofficial-vendor-reference\u002Fmicrosoft-azure-webjobs-extensions-authentication-events-dotnet\u002Freferences\u002Facceptance-criteria.md",[94,123,157,184,210,227],{"skillNo":95,"name":96,"nameZh":6,"description":97,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":98,"categories":99,"keywords":109,"publishedAt":44},"01M6GJTE1GESGGF9K028Y7753E","xurl","X\u002FTwitter via xurl CLI: post, search, DM, media, v2 API.","https:\u002F\u002Fgithub.com\u002Fzapabob\u002FSkills\u002Fblob\u002Fmain\u002Fregistry\u002Fskills\u002Fxurl\u002Fvariants\u002Fhermes-agent\u002FSKILL.md",[100,103,104],{"categoryNo":101,"categoryName":102,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000010","后端与接口",{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},{"categoryNo":105,"categoryName":106,"taxonomyType":107,"taxonomyTypeDesc":108},"01JBSEEDCAT000000000000030","开发工程师","PROFESSION","适用职业",[96,58,64,110,111,43,39,112,113,114,115,116,117,32,118,119,65,120,121,122],"cli","github","official","search","https","social-media","api","Media","hermes agent","twitter","social","via","post",{"skillNo":124,"name":125,"nameZh":6,"description":126,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":127,"categories":128,"keywords":140,"publishedAt":44},"01M6GJT8S1PXQ8WPXT1HSNBWH2","writing-plans","Write implementation plans: bite-sized tasks, paths, code.","https:\u002F\u002Fgithub.com\u002Fzapabob\u002FSkills\u002Fblob\u002Fmain\u002Fregistry\u002Fskills\u002Fwriting-plans\u002Fvariants\u002Fhermes-agent\u002FSKILL.md",[129,132,135,136,137],{"categoryNo":130,"categoryName":131,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000015","视觉设计",{"categoryNo":133,"categoryName":134,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000018","写作与文案",{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},{"categoryNo":105,"categoryName":106,"taxonomyType":107,"taxonomyTypeDesc":108},{"categoryNo":138,"categoryName":139,"taxonomyType":107,"taxonomyTypeDesc":108},"01JBSEEDCAT000000000000036","内容创作者",[141,142,143,35,64,144,43,145,39,146,147,148,149,150,151,152,153,154,155,156],"writing","writing plans","plans","review","workflow","documentation","tasks","development","code","planning","code-review","subagent","write","superpowers","paths","test-driven-development",{"skillNo":158,"name":159,"nameZh":6,"description":160,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":161,"categories":162,"keywords":166,"publishedAt":44},"01M6GJST0H5X7WHBJ3BZK0DVKH","whisper","OpenAI's general-purpose speech recognition model. Supports 99 languages, transcription, translation to English, and language identification. Six model sizes from tiny (39M params) to large (1550M params). Use for speech-to-text, podcast transcription, or multilingual audio processing. Best for robust, multilingual ASR.","https:\u002F\u002Fgithub.com\u002Fzapabob\u002FSkills\u002Fblob\u002Fmain\u002Fregistry\u002Fskills\u002Fwhisper\u002Fvariants\u002Fhermes-agent-optional\u002FSKILL.md",[163],{"categoryNo":164,"categoryName":165,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000016","图像与多媒体",[159,64,43,167,168,169,170,171,172,173,174,175,176,177,178,179,180,181,182,183],"openai","speech","multimodal","Transformers","ASR","Audio","text","model","speech recognition","use","general","multilingual","translation","english","identification","processing","Audio Processing",{"skillNo":185,"name":186,"nameZh":6,"description":187,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":188,"categories":189,"keywords":193,"publishedAt":44},"01M6GJSMRQ8EETZTYMG72P70EA","weights-and-biases","W&B: log ML experiments, sweeps, model registry, dashboards.","https:\u002F\u002Fgithub.com\u002Fzapabob\u002FSkills\u002Fblob\u002Fmain\u002Fregistry\u002Fskills\u002Fweights-and-biases\u002Fvariants\u002Fhermes-agent\u002FSKILL.md",[190],{"categoryNo":191,"categoryName":192,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000013","分析与可视化",[194,64,195,196,39,197,198,199,200,201,202,174,203,31,204,205,206,207,208,209],"weights and biases","tensorflow","experiments","pytorch","visualization","registry","real-time","tuning","dashboards","experiment","time","Experiment Tracking","Model Registry","Hyperparameter Tuning","WandB","Real-Time Visualization",{"skillNo":211,"name":212,"nameZh":6,"description":213,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":214,"categories":215,"keywords":217,"publishedAt":44},"01M6GJS19KDPRY112JTR326S06","watchers","Poll RSS, JSON APIs, and GitHub with watermark dedup.","https:\u002F\u002Fgithub.com\u002Fzapabob\u002FSkills\u002Fblob\u002Fmain\u002Fregistry\u002Fskills\u002Fwatchers\u002Fvariants\u002Fhermes-agent-optional\u002FSKILL.md",[216],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},[212,64,218,111,43,219,220,221,222,223,224,32,118,225,226],"terminal","automation","http","monitoring","cron","json","rss","devops","with",{"skillNo":228,"name":229,"nameZh":6,"description":230,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":231,"categories":232,"keywords":235,"publishedAt":44},"01M6GJRHJD2A6Q830SWF9W48BC","serving-llms-vllm","vLLM: high-throughput LLM serving, OpenAI API, quantization.","https:\u002F\u002Fgithub.com\u002Fzapabob\u002FSkills\u002Fblob\u002Fmain\u002Fregistry\u002Fskills\u002Fvllm\u002Fvariants\u002Fhermes-agent\u002FSKILL.md",[233,234],{"categoryNo":101,"categoryName":102,"taxonomyType":21,"taxonomyTypeDesc":22},{"categoryNo":105,"categoryName":106,"taxonomyType":107,"taxonomyTypeDesc":108},[236,237,238,43,239,167,240,241,242,170,243,116,244,245,246,247,248,249,250,251],"llms","vLLM","serving llms vllm","research","openai api","tensor","Inference Serving","Production","llm serving","High Throughput","PagedAttention","Continuous Batching","Quantization","Tensor Parallelism","high-throughput","throughput"]