[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M69B0NBHGZRRZBGP5CHPBH91":3,"skill-same:01M69B0NBHGZRRZBGP5CHPBH91":54},{"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":26,"linkOnly":27,"markdownContent":28,"markdownBody":29,"frontMatter":30,"refName":31,"commitSha":32,"skillDirectory":33,"forksCount":34,"repositoryHtmlUrl":35,"repositoryHomepageUrl":6,"ownerHtmlUrl":36,"ownerBlogUrl":37,"ownerDescription":38,"updatedAt":39,"install":40,"resources":53},"01M69B0NBHGZRRZBGP5CHPBH91","create-demo-video",null,"Produce a narrated demo video that shows something the user made or built, cut from recordings they supply or recordings made to their direction. Use when showing a product, feature, tool, or project in action, including explainers built around a live demo. Not for slide-style explainers; for a single trim or audio cleanup, use the relevant tool directly.","microsoft\u002Famplifier-bundle-skills","microsoft","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F6154722?v=4","01M698VYDXZ6JW5WH3CRV650R8","Microsoft","01M69AZYFW50CECTV9VW6B34SE",12,"MIT","https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Famplifier-bundle-skills\u002Fblob\u002Fmain\u002Fskills\u002Fcreate-demo-video\u002FSKILL.md",[18],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000016","图像与多媒体","SCENARIO","使用场景",[24,25],"demo video","create demo video","2026-11-10 10:12:30",false,"---\nname: create-demo-video\ndescription: >-\n  Produce a narrated demo video that shows something the user made or built,\n  cut from recordings they supply or recordings made to their direction. Use\n  when showing a product, feature, tool, or project in action, including\n  explainers built around a live demo. Not for slide-style explainers; for a\n  single trim or audio cleanup, use the relevant tool directly.\n---\n\n# Create a Demo Video\n\nShow the thing working. Build every demo scene from a recording of it, never\nfrom slides or static screenshots standing in for it. Settle the storyboard and\nnarration before detailed visual timing, then review one canonical export end\nto end.\n\n## Harness\n\nDrive this workflow from a harness running a strong reasoning model, such as\nthe Amplifier CLI with `gpt-6-astra` at high reasoning effort or Claude Opus 5.5.\nThe workflow depends on sustained judgment across footage, narration timing,\nand review.\n\n## Tools\n\nInstall the `smart-tools` skill from\n[microsoft\u002Famplifier-smart-tools](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Famplifier-smart-tools):\n\n```bash\nnpx skills add microsoft\u002Famplifier-smart-tools\n```\n\nFor the Amplifier app CLI, add the Smart Tools behavior instead:\n\n```bash\namplifier bundle add 'git+https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Famplifier-smart-tools@main#subdirectory=behaviors\u002Fsmart-tools.yaml' --app\n```\n\nLoad it to access the catalog and each tool's installation, prerequisites, and\nusage guidance:\n\n- **vid:** edit plans, assembly, rendering, and export verification.\n- **aud:** speech cleanup, mastering, and loudness\u002Fpeak verification.\n- **unfold:** silent animations with editable source.\n- **showrun:** recorded walkthroughs of a running web app or terminal.\n- **Stories:** storyboards and speech synthesis. Prefer Gemini 3.1 TTS\n  (`gemini-3.1-flash-tts-preview`); audition a short passage to choose the voice.\n  Master the resulting speech with aud.\n\nCall the smart tool that owns each step: Stories for the storyboard and\nnarration, Showrun for recordings, aud for mastering, Unfold for animation, and\nvid for video edits. Do not reimplement a tool's work with hand-written commands\nor scripts; use scripts only to chain smart tool capabilities into repeatable\nbuilds.\n\n## Workflow\n\nBefore step 1, run `--help` for each smart tool the video will use: `stories`,\n`showrun`, `aud`, `unfold`, and `vid`. Follow what the help says about setup,\ncapabilities, and arguments in every later step.\n\n### 1. Create the storyboard with Stories\n\nInspect existing scripts, recordings, artwork, and credits. Establish the\naudience, takeaway, runtime limit, and destination; ask only for missing\ndecisions. Use 1080p at 30 fps unless the brief calls for another format.\n\nStart with a hook, a tour of visible results, a short synthesis, and a closing\naction. Give each demo one spoken point and footage that proves it. Reuse chosen\nbranding; otherwise compare a few small layout previews and voice samples\nbefore applying a direction throughout.\n\nBuild the storyboard in Stories. Use `create-storyboard` to import a script the\nuser already has, or `generate-storyboard` to develop one from the brief. Set\n`explore` only when the user asks to compare directions. Each panel is one\nscene:\n\n- `id`: stable scene ID.\n- `action`: the point the scene makes.\n- `visual`: what is on screen: a recording excerpt with file and in\u002Fout times,\n  an Unfold animation such as an opening title or concept diagram, a title\n  card, or a combination.\n- `narration`: the spoken line.\n- `notes` (optional): on-screen text, such as titles, animation labels, and\n  subtitles.\n- `production_requirements` (optional): recordings or animations still to make.\n- `asset_id` (optional): a still frame from the recording.\n\nA direction holds up to eight panels; for a longer video, make each panel a\nsection and list its scenes in `visual` and `narration`. Share the storyboard\nthrough the Stories dashboard with a short voice sample, and wait for the user\nto approve both before recording or generating narration, unless the user has\nsaid to proceed without review. The storyboard is the source of truth for every\nlater step. Apply changes with `revise-storyboard` and rebuild from the latest\nrevision.\n\nUse the scene IDs to connect on-screen sources, narration, and review\ntimestamps. Preserve original assets, animation source, and generation\nsettings; separate caches from rebuild inputs.\n\n**Success criteria:** Each scene has a purpose, an on-screen source, and a\nnarration line, the total fits the runtime budget, and the user has approved the\nstoryboard and voice or waived review.\n\n### 2. Get the recordings\n\nUse recordings the user supplies, or make recordings of what the user wants to\nshow. Long, unedited captures are fine; step 4 cuts them down. Make recordings\nwith real screen capture of the thing running, using Showrun for web apps and\nits terminal mode for command-line demos. Seed realistic demo data first so\nscreens do not start empty. Record a trial take before the real one, especially\nwhen a take spends money or time. Start each take with the screen and scrollback\ncleared, keep terminal takes to one or two commands, and check the receipt that\nevery step shows a first interaction. Use Showrun's receipts, which time each\naction and wait, to choose cut points. Share takes with the user through\nShowrun's review dashboard before cutting them. Only substitute simulated or\nanimated footage when the user explicitly asks not to use real screen capture.\nIf a storyboard beat cannot be recorded because the thing lacks it, tell the\nuser rather than faking it.\n\n**Success criteria:** Every demo scene in the storyboard maps to a recording\nthat shows it happening.\n\n### 3. Finish narration before timing visuals\n\nClean supplied speech or generate one clip per scene with consistent voice\nsettings. Check the speech provider's quota before auditioning voices;\nauditions spend the same budget as the final narration. Generate all lines with\none provider so the voice stays consistent. Use aud to master the clips; -16\nLUFS and a -1.5 dBTP ceiling are useful starting points. Inspect the\nverification results, and leave judgments of voice quality to the person\nreviewing in step 7. Retain raw and mastered clips; regenerate only changed\nlines.\n\nMeasure speech and set scene durations with breathing room. If the video runs\nlong, shorten copy and remove redundant beats before speeding up speech or demos.\nKeep spoken copy and on-screen claims synchronized. Regenerated speech requires\nnew timing cues, including any timestamp-based breath or pause edits.\n\n**Success criteria:** Narration sounds consistent, fits each scene without\nclipping, and leaves the complete video within its runtime budget.\n\n### 4. Cut demos around the result\n\nCut with vid's trim, cut, and retime verbs. Show enough setup to explain the\naction, meaningful progress, and a readable result. Remove waiting and\nrepetition; accelerate only where viewers can still follow. Judge the excerpt at\ndelivery size and speed.\n\nPreserve the full recording and aspect ratio unless a deliberate crop is wanted.\nWhere vid supports it, introduce the name, author, and description beside the\nrecording, then expand it smoothly to full screen while playback continues. A\n2.5-second introduction and 0.9-second expansion are a useful starting point.\nOtherwise open the scene with a title card. Hold final outputs long enough to\nunderstand them.\n\n**Success criteria:** Every excerpt supports its spoken point without hiding\ncontent or rushing past the useful result.\n\n### 5. Animate to the measured narration\n\nGive Unfold the meaning, exact labels, palette, canvas size, frame rate, duration,\nreveal timestamps, and final hold. Request silent output. For explanations, build\na stable diagram that reveals concepts in spoken order. Check a short prototype\nbefore rendering the full sequence. Unfold `create` runs for minutes; run\ncreates one at a time with a timeout long enough to finish, not as background\njobs the harness may kill.\n\nRetain editable source and dependencies. Make narrow text, alignment, and timing\nchanges there instead of regenerating the animation. Recompute cues when speech\nchanges. Inspect encoded frames at reveals, transitions, and the ending for\ncollisions, misleading arrows, small text, and blank frames. Share animations\nwith the user through Unfold's dashboard and apply their notes with Unfold's\nrefinement before assembly.\n\n**Success criteria:** The animation matches the narration, remains readable,\nand holds a complete final composition.\n\n### 6. Assemble and mix\n\nBuild the whole video as one vid pipeline: stitch the scenes with vid\ntransitions, lay the narration in with vid's audio verbs, and render once.\nSave the plan from `vid plan` with the project so the edit can be reviewed and\nreplayed.\nAccount for transition overlaps so later narration does not drift; extra\noutgoing footage can compensate for an overlap. A dissolve crossfades audio as\nwell as picture, so start each scene's narration after its incoming transition\nends, or use a hard cut where a line must start immediately. Keep audio within\nthe picture's duration without cutting speech.\n\nMute recording audio only where voiceover replaces it. Where recorded and\nsynthetic voices meet, add the handoff to the checks for a person in step 7.\n\n**Success criteria:** The export has the intended runtime, dimensions, frame\nrate, and audio, with aligned transitions and intelligible speech.\n\n### 7. Review and revise the complete video\n\nRun the tool checks yourself: `vid verify` on the render, `aud verify` on the\nfinal mix, a transcription of the mix compared against the storyboard\nnarration, and frames sampled at every scene boundary and in the final seconds.\nLook for black flashes, clipped speech, frozen endings, and abrupt cutoffs. On\ndark palettes, `vid verify --expect-no-black-frames` can flag designed frames;\nconfirm with the sampled frames before treating them as a defect. If `aud verify`\nmisses the loudness target, master the extracted mix with aud, swap it in with\n`vid audio replace`, and render again; this is the one expected second render.\nModel reviews of picture or sound do not replace a person. List the checks only\na person can make, such as pacing, voice quality, and whether the demo is\nconvincing, as open items in the delivery report.\n\nKeep one canonical video for feedback, addressed by scene or timestamp. Revise\nthe saved vid plan and rebuild only affected assets before rendering the full\nexport again.\n\nCreate a local HTML review player for the canonical MP4. Include:\n\n- A video player with playback, seeking, volume, and fullscreen controls.\n- Selectable closed captions, enabled by default when captions are requested.\n  Build the caption file from the storyboard narration and the scene timeline.\n- Scene buttons labeled with each scene's start time and title from the\n  storyboard. Clicking a button seeks to that scene and starts playback.\n- Links to download the MP4 and caption file, and to view the storyboard.\n- A responsive layout using the video's branding, with the player above the\n  scene buttons.\n\nSave the player as `index.html` alongside the video and caption files. Serve it\non localhost and open it for review. Start paused so the user controls playback.\nKeep one canonical video and update the player's scene timings whenever the edit\nchanges.\n\nVerify that the video plays, captions display correctly, and every scene button\nseeks to the intended moment. Deliver the MP4 as the primary artifact and link\nthe review player as the feedback surface. The review player is supporting HTML\ncreated by the coordinating agent; it does not need to come from a smart tool.\n\nKeep review aids, the storyboard, original assets, animation sources, and\nrebuild commands with prerequisites in the project rather than bundling them\ninto an archive. Report what was actually verified and any viewing or listening\nchecks that still need the user.\n\n**Success criteria:** The complete video meets the brief and can be revised\nwithout reconstructing the project.\n","# Create a Demo Video\n\nShow the thing working. Build every demo scene from a recording of it, never\nfrom slides or static screenshots standing in for it. Settle the storyboard and\nnarration before detailed visual timing, then review one canonical export end\nto end.\n\n## Harness\n\nDrive this workflow from a harness running a strong reasoning model, such as\nthe Amplifier CLI with `gpt-6-astra` at high reasoning effort or Claude Opus 5.5.\nThe workflow depends on sustained judgment across footage, narration timing,\nand review.\n\n## Tools\n\nInstall the `smart-tools` skill from\n[microsoft\u002Famplifier-smart-tools](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Famplifier-smart-tools):\n\n```bash\nnpx skills add microsoft\u002Famplifier-smart-tools\n```\n\nFor the Amplifier app CLI, add the Smart Tools behavior instead:\n\n```bash\namplifier bundle add 'git+https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Famplifier-smart-tools@main#subdirectory=behaviors\u002Fsmart-tools.yaml' --app\n```\n\nLoad it to access the catalog and each tool's installation, prerequisites, and\nusage guidance:\n\n- **vid:** edit plans, assembly, rendering, and export verification.\n- **aud:** speech cleanup, mastering, and loudness\u002Fpeak verification.\n- **unfold:** silent animations with editable source.\n- **showrun:** recorded walkthroughs of a running web app or terminal.\n- **Stories:** storyboards and speech synthesis. Prefer Gemini 3.1 TTS\n  (`gemini-3.1-flash-tts-preview`); audition a short passage to choose the voice.\n  Master the resulting speech with aud.\n\nCall the smart tool that owns each step: Stories for the storyboard and\nnarration, Showrun for recordings, aud for mastering, Unfold for animation, and\nvid for video edits. Do not reimplement a tool's work with hand-written commands\nor scripts; use scripts only to chain smart tool capabilities into repeatable\nbuilds.\n\n## Workflow\n\nBefore step 1, run `--help` for each smart tool the video will use: `stories`,\n`showrun`, `aud`, `unfold`, and `vid`. Follow what the help says about setup,\ncapabilities, and arguments in every later step.\n\n### 1. Create the storyboard with Stories\n\nInspect existing scripts, recordings, artwork, and credits. Establish the\naudience, takeaway, runtime limit, and destination; ask only for missing\ndecisions. Use 1080p at 30 fps unless the brief calls for another format.\n\nStart with a hook, a tour of visible results, a short synthesis, and a closing\naction. Give each demo one spoken point and footage that proves it. Reuse chosen\nbranding; otherwise compare a few small layout previews and voice samples\nbefore applying a direction throughout.\n\nBuild the storyboard in Stories. Use `create-storyboard` to import a script the\nuser already has, or `generate-storyboard` to develop one from the brief. Set\n`explore` only when the user asks to compare directions. Each panel is one\nscene:\n\n- `id`: stable scene ID.\n- `action`: the point the scene makes.\n- `visual`: what is on screen: a recording excerpt with file and in\u002Fout times,\n  an Unfold animation such as an opening title or concept diagram, a title\n  card, or a combination.\n- `narration`: the spoken line.\n- `notes` (optional): on-screen text, such as titles, animation labels, and\n  subtitles.\n- `production_requirements` (optional): recordings or animations still to make.\n- `asset_id` (optional): a still frame from the recording.\n\nA direction holds up to eight panels; for a longer video, make each panel a\nsection and list its scenes in `visual` and `narration`. Share the storyboard\nthrough the Stories dashboard with a short voice sample, and wait for the user\nto approve both before recording or generating narration, unless the user has\nsaid to proceed without review. The storyboard is the source of truth for every\nlater step. Apply changes with `revise-storyboard` and rebuild from the latest\nrevision.\n\nUse the scene IDs to connect on-screen sources, narration, and review\ntimestamps. Preserve original assets, animation source, and generation\nsettings; separate caches from rebuild inputs.\n\n**Success criteria:** Each scene has a purpose, an on-screen source, and a\nnarration line, the total fits the runtime budget, and the user has approved the\nstoryboard and voice or waived review.\n\n### 2. Get the recordings\n\nUse recordings the user supplies, or make recordings of what the user wants to\nshow. Long, unedited captures are fine; step 4 cuts them down. Make recordings\nwith real screen capture of the thing running, using Showrun for web apps and\nits terminal mode for command-line demos. Seed realistic demo data first so\nscreens do not start empty. Record a trial take before the real one, especially\nwhen a take spends money or time. Start each take with the screen and scrollback\ncleared, keep terminal takes to one or two commands, and check the receipt that\nevery step shows a first interaction. Use Showrun's receipts, which time each\naction and wait, to choose cut points. Share takes with the user through\nShowrun's review dashboard before cutting them. Only substitute simulated or\nanimated footage when the user explicitly asks not to use real screen capture.\nIf a storyboard beat cannot be recorded because the thing lacks it, tell the\nuser rather than faking it.\n\n**Success criteria:** Every demo scene in the storyboard maps to a recording\nthat shows it happening.\n\n### 3. Finish narration before timing visuals\n\nClean supplied speech or generate one clip per scene with consistent voice\nsettings. Check the speech provider's quota before auditioning voices;\nauditions spend the same budget as the final narration. Generate all lines with\none provider so the voice stays consistent. Use aud to master the clips; -16\nLUFS and a -1.5 dBTP ceiling are useful starting points. Inspect the\nverification results, and leave judgments of voice quality to the person\nreviewing in step 7. Retain raw and mastered clips; regenerate only changed\nlines.\n\nMeasure speech and set scene durations with breathing room. If the video runs\nlong, shorten copy and remove redundant beats before speeding up speech or demos.\nKeep spoken copy and on-screen claims synchronized. Regenerated speech requires\nnew timing cues, including any timestamp-based breath or pause edits.\n\n**Success criteria:** Narration sounds consistent, fits each scene without\nclipping, and leaves the complete video within its runtime budget.\n\n### 4. Cut demos around the result\n\nCut with vid's trim, cut, and retime verbs. Show enough setup to explain the\naction, meaningful progress, and a readable result. Remove waiting and\nrepetition; accelerate only where viewers can still follow. Judge the excerpt at\ndelivery size and speed.\n\nPreserve the full recording and aspect ratio unless a deliberate crop is wanted.\nWhere vid supports it, introduce the name, author, and description beside the\nrecording, then expand it smoothly to full screen while playback continues. A\n2.5-second introduction and 0.9-second expansion are a useful starting point.\nOtherwise open the scene with a title card. Hold final outputs long enough to\nunderstand them.\n\n**Success criteria:** Every excerpt supports its spoken point without hiding\ncontent or rushing past the useful result.\n\n### 5. Animate to the measured narration\n\nGive Unfold the meaning, exact labels, palette, canvas size, frame rate, duration,\nreveal timestamps, and final hold. Request silent output. For explanations, build\na stable diagram that reveals concepts in spoken order. Check a short prototype\nbefore rendering the full sequence. Unfold `create` runs for minutes; run\ncreates one at a time with a timeout long enough to finish, not as background\njobs the harness may kill.\n\nRetain editable source and dependencies. Make narrow text, alignment, and timing\nchanges there instead of regenerating the animation. Recompute cues when speech\nchanges. Inspect encoded frames at reveals, transitions, and the ending for\ncollisions, misleading arrows, small text, and blank frames. Share animations\nwith the user through Unfold's dashboard and apply their notes with Unfold's\nrefinement before assembly.\n\n**Success criteria:** The animation matches the narration, remains readable,\nand holds a complete final composition.\n\n### 6. Assemble and mix\n\nBuild the whole video as one vid pipeline: stitch the scenes with vid\ntransitions, lay the narration in with vid's audio verbs, and render once.\nSave the plan from `vid plan` with the project so the edit can be reviewed and\nreplayed.\nAccount for transition overlaps so later narration does not drift; extra\noutgoing footage can compensate for an overlap. A dissolve crossfades audio as\nwell as picture, so start each scene's narration after its incoming transition\nends, or use a hard cut where a line must start immediately. Keep audio within\nthe picture's duration without cutting speech.\n\nMute recording audio only where voiceover replaces it. Where recorded and\nsynthetic voices meet, add the handoff to the checks for a person in step 7.\n\n**Success criteria:** The export has the intended runtime, dimensions, frame\nrate, and audio, with aligned transitions and intelligible speech.\n\n### 7. Review and revise the complete video\n\nRun the tool checks yourself: `vid verify` on the render, `aud verify` on the\nfinal mix, a transcription of the mix compared against the storyboard\nnarration, and frames sampled at every scene boundary and in the final seconds.\nLook for black flashes, clipped speech, frozen endings, and abrupt cutoffs. On\ndark palettes, `vid verify --expect-no-black-frames` can flag designed frames;\nconfirm with the sampled frames before treating them as a defect. If `aud verify`\nmisses the loudness target, master the extracted mix with aud, swap it in with\n`vid audio replace`, and render again; this is the one expected second render.\nModel reviews of picture or sound do not replace a person. List the checks only\na person can make, such as pacing, voice quality, and whether the demo is\nconvincing, as open items in the delivery report.\n\nKeep one canonical video for feedback, addressed by scene or timestamp. Revise\nthe saved vid plan and rebuild only affected assets before rendering the full\nexport again.\n\nCreate a local HTML review player for the canonical MP4. Include:\n\n- A video player with playback, seeking, volume, and fullscreen controls.\n- Selectable closed captions, enabled by default when captions are requested.\n  Build the caption file from the storyboard narration and the scene timeline.\n- Scene buttons labeled with each scene's start time and title from the\n  storyboard. Clicking a button seeks to that scene and starts playback.\n- Links to download the MP4 and caption file, and to view the storyboard.\n- A responsive layout using the video's branding, with the player above the\n  scene buttons.\n\nSave the player as `index.html` alongside the video and caption files. Serve it\non localhost and open it for review. Start paused so the user controls playback.\nKeep one canonical video and update the player's scene timings whenever the edit\nchanges.\n\nVerify that the video plays, captions display correctly, and every scene button\nseeks to the intended moment. Deliver the MP4 as the primary artifact and link\nthe review player as the feedback surface. The review player is supporting HTML\ncreated by the coordinating agent; it does not need to come from a smart tool.\n\nKeep review aids, the storyboard, original assets, animation sources, and\nrebuild commands with prerequisites in the project rather than bundling them\ninto an archive. Report what was actually verified and any viewing or listening\nchecks that still need the user.\n\n**Success criteria:** The complete video meets the brief and can be revised\nwithout reconstructing the project.\n",{"name":5,"description":7},"main","2f58408a879535df36da5c92d2b073f49271ade4","skills\u002Fcreate-demo-video",11,"https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Famplifier-bundle-skills","https:\u002F\u002Fgithub.com\u002Fmicrosoft","https:\u002F\u002Fopensource.microsoft.com","Open source projects and samples from Microsoft","2026-11-12 16:43:43",{"source":8,"skillName":5,"commands":41,"prompt":51,"zipAvailable":52},[42,45,48],{"tool":43,"command":44},"npx","npx skills add microsoft\u002Famplifier-bundle-skills --skill create-demo-video",{"tool":46,"command":47},"bunx","bunx skills add microsoft\u002Famplifier-bundle-skills --skill create-demo-video",{"tool":49,"command":50},"pnpm","pnpm dlx skills add microsoft\u002Famplifier-bundle-skills --skill create-demo-video","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Famplifier-bundle-skills\u002Fblob\u002Fmain\u002Fskills\u002Fcreate-demo-video\u002FSKILL.md",true,[],[55,65,76,89,100,112],{"skillNo":56,"name":57,"nameZh":6,"description":58,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":59,"categories":60,"keywords":61,"publishedAt":26},"01M69B282M59QZNMM26JXA1QCV","verification-discipline","Use when verifying that completed work actually works. Auto-surface during \u002Fverify mode, post-implementation review, or before claiming a task is done. Teaches testing outcomes vs implementation, the unit\u002Fintegration\u002Fsmoke gradient, and what \"done\" actually means.","https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Famplifier-bundle-skills\u002Fblob\u002Fmain\u002Fskills\u002Fverification-discipline\u002FSKILL.md",[],[62,63,64],"verification discipline","review","testing",{"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":26},"01M69B263M6BFCVAV3SEESV7T0","user-advocate","User-need reviewer that speaks for the person who isn't in the room — the one who will actually live with what gets built. Hunts the gap between \"we can build this\" and \"they actually want this,\" in the patient, impatient voice of the absent user. Use at any checkpoint — brainstorm, design, plan, implement, debug, review — when a feature is being built because it's buildable rather than wanted, the recovery path is missing, or nobody can name the person this serves: \"do they actually want this, and can they live with it?\"","https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Famplifier-bundle-skills\u002Fblob\u002Fmain\u002Fskills\u002Fuser-advocate\u002FSKILL.md",[],[72,73,74,63,75],"user advocate","debug","implement","design",{"skillNo":77,"name":78,"nameZh":6,"description":79,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":80,"categories":81,"keywords":87,"publishedAt":26},"01M69B23EQQM30TMZG7FPJW4F7","tester-breaker","Adversarial breaker that reviews code by trying to make it fail, not by confirming it works. Hunts the unhappy paths — malformed input, the empty string, the reversed range, the race condition — as a gleeful adversary who assumes everything is broken until a concrete attempt to break it comes up empty. Use at any checkpoint — brainstorm, design, plan, implement, debug, review — when the happy path is celebrated while the edges sit unexamined, \"looks fine\" stands in for \"I tried to break it and couldn't,\" or nobody has named the input that makes this fall over: \"how does this fail, and where are the edges?\"","https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Famplifier-bundle-skills\u002Fblob\u002Fmain\u002Fskills\u002Ftester-breaker\u002FSKILL.md",[82],{"categoryNo":83,"categoryName":84,"taxonomyType":85,"taxonomyTypeDesc":86},"01JBSEEDCAT000000000000030","开发工程师","PROFESSION","适用职业",[88,63,73,74,75],"tester breaker",{"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":26},"01M69B21AHBXP6EK60XYDP24PY","skills-assist","Authoritative consultant for all skills-related questions. Use when creating or modifying skills, understanding the Agent Skills spec, troubleshooting skill loading or invocation issues, leveraging enhanced format features (context fork, model_role, user-invocable), writing cross-harness portable skills, ensuring Claude Code Skills 2.0 compatibility, or deciding between skills vs agents.","https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Famplifier-bundle-skills\u002Fblob\u002Fmain\u002Fskills\u002Fskills-assist\u002FSKILL.md",[],[96,97,98,99],"skills assist","claude code skills","spec","writing",{"skillNo":101,"name":102,"nameZh":6,"description":103,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":104,"categories":105,"keywords":109,"publishedAt":26},"01M69B1ZBHGE1PAX0E7ZDS8HQZ","skillify","Capture a repeatable process from the current session into a reusable Amplifier SKILL.md skill file. Use when the user wants to create a skill, save a workflow as a skill, turn a process into a reusable skill, or mentions \"skillify\", \"create skill\", \"make a skill\", \"save as skill\", \"capture workflow\", \"turn this into a skill\", \"new skill\", or wants to automate a repeatable process they just performed.","https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Famplifier-bundle-skills\u002Fblob\u002Fmain\u002Fskills\u002Fskillify\u002FSKILL.md",[106],{"categoryNo":107,"categoryName":108,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000023","Skill与智能体开发",[102,110,111],"create skill","workflow",{"skillNo":113,"name":114,"nameZh":6,"description":115,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":116,"categories":117,"keywords":119,"publishedAt":26},"01M69B1XCPQFAXS9RTS87CZ8EK","session-debug","Diagnose issues in the current Amplifier session — misconfigured tools, failing operations, unexpected behavior. Use when something isn't working right.","https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Famplifier-bundle-skills\u002Fblob\u002Fmain\u002Fskills\u002Fsession-debug\u002FSKILL.md",[118],{"categoryNo":83,"categoryName":84,"taxonomyType":85,"taxonomyTypeDesc":86},[120,73,121],"session debug","diagnose"]