[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M71JD1GZ38JWA31DD0M2KN41":3,"skill-same:01M71JD1GZ38JWA31DD0M2KN41":111},{"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":43,"linkOnly":44,"markdownContent":45,"markdownBody":46,"frontMatter":47,"refName":48,"commitSha":49,"skillDirectory":5,"forksCount":50,"repositoryHtmlUrl":51,"repositoryHomepageUrl":52,"ownerHtmlUrl":53,"ownerBlogUrl":54,"ownerDescription":55,"updatedAt":56,"install":57,"resources":70},"01M71JD1GZ38JWA31DD0M2KN41","clipsmith",null,"Plan and run a custom video-editing pipeline from a plain-language brief — clip, cut, trim, montage, reframe, or caption any footage. Use this whenever the user wants to edit or auto-clip video: make a highlight reel, a montage, a supercut, a beat-synced edit, a vertical\u002FTikTok\u002FReels\u002FShorts clip, remove silences or dead air, do jump-cuts, add captions\u002Fsubtitles, mix or duck music under footage, speed-ramp, or pull the best \u002F most interesting moments out of one or more videos — even when they don't say \"edit\" and just say things like \"turn these clips into something postable\", \"find the good bits\", \"make this shorter\", or \"cut me a reel\". Composes ffmpeg with audio \u002F scene \u002F motion analysis, optional speech transcription, and the agent's OWN VISION for judging which moments are actually good. Not domain-specific — works on any footage (sports, vlogs, gameplay, talking-head, b-roll).","caezium\u002Fskills","caezium","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F113233555?v=4","01M71JB77ZPBAWEK009Y2W8BP7","Henry Zhang","01M71JB792NVXCWSYD27HQ0964",3,"MIT","https:\u002F\u002Fgithub.com\u002Fcaezium\u002Fskills\u002Fblob\u002Fmain\u002Fclipsmith\u002FSKILL.md",[18],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000016","图像与多媒体","SCENARIO","使用场景",[5,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42],"tiktok","me","plan","analysis","when","synced","auto","trim","good","remove","mix","pipeline","Captions","Audio","reel","which","reframe","brief","Video","2026-11-10 10:12:30",false,"---\nname: clipsmith\ndescription: >-\n  Plan and run a custom video-editing pipeline from a plain-language brief — clip, cut, trim,\n  montage, reframe, or caption any footage. Use this whenever the user wants to edit or\n  auto-clip video: make a highlight reel, a montage, a supercut, a beat-synced edit, a\n  vertical\u002FTikTok\u002FReels\u002FShorts clip, remove silences or dead air, do jump-cuts, add\n  captions\u002Fsubtitles, mix or duck music under footage, speed-ramp, or pull the best \u002F most\n  interesting moments out of one or more videos — even when they don't say \"edit\" and just\n  say things like \"turn these clips into something postable\", \"find the good bits\", \"make this\n  shorter\", or \"cut me a reel\". Composes ffmpeg with audio \u002F scene \u002F motion analysis, optional\n  speech transcription, and the agent's OWN VISION for judging which moments are actually good.\n  Not domain-specific — works on any footage (sports, vlogs, gameplay, talking-head, b-roll).\n---\n\n# clipsmith\n\nYou are an automatic video editor. The user gives you footage and a rough brief; you compose\nand run a **custom pipeline** for that specific request, then verify the result actually matches\nwhat they asked for. There is no fixed template — every brief implies a slightly different graph\nof operations. Your job is to pick the right operations and wire them together.\n\nThe toolkit is general (ffmpeg + a few analysis scripts), so resist the urge to hand-write\nfragile one-off ffmpeg incantations for the fiddly parts — the bundled scripts already handle\nthe things that are easy to get subtly wrong (concat path resolution, uniform re-encoding,\nreframing, loudness). Spend your effort on the *editorial* decisions instead.\n\n## The loop\n\nWork through these phases. Skip what a brief doesn't need (e.g. a \"remove the silences\" job\nneeds no vision step; a \"best moments\" job leans on it heavily), but never skip **probe** or\n**verify**.\n\n1. **Read the brief → a target.** Pin down: output length, aspect ratio (vertical 9:16 \u002F\n   square \u002F landscape), whether there's music, captions yes\u002Fno, and the *vibe* (punchy and\n   fast vs. calm and long). If two or three of these are genuinely ambiguous and would change\n   the result a lot, ask one tight question; otherwise pick sensible defaults and say what you\n   chose. Don't interrogate.\n\n2. **Probe the media.** `scripts\u002Fprobe.py FILE...` → JSON of duration, fps, resolution, audio.\n   You need this before planning anything (a 9:16 source needs no cropping; a silent clip can't\n   be silence-trimmed; a 4-minute source needs a proxy).\n\n3. **Proxy if heavy.** For anything long or high-res, make a 720p proxy first and do all\n   *analysis* on it (`scripts\u002Fproxy.py`). Render the final cut from the originals. Analysis on\n   full 4K is slow and pointless.\n\n4. **Analyze.** `scripts\u002Fanalyze.py FILE` computes, per source, a per-second \"interestingness\"\n   score (audio energy + onset + brightness, motion, scene-cut bonus) plus the music beat grid,\n   scene-cut times, and silence spans. It caches to a sidecar `.clipsmith.json`, so re-running\n   is free. Pick the signals the brief needs via flags. This finds *activity*.\n\n5. **Judge with your own eyes (the taste step).** Activity ≠ interesting. This is what separates\n   a good edit from a motion-energy slideshow. Pull candidate frames at the top peaks\n   (`scripts\u002Fframes.py FILE --at \u003Ctimes>` or `--from-analysis`) into a contact sheet, **look at\n   them**, and rank which moments actually fit the brief (\"overtake\", \"the trick lands\", \"she\n   laughs\", \"clean wide shot\"). Drop the ones that scored high on noise but look boring. For\n   longer candidates, sample 2–3 frames across the span. Your visual judgment is the feature\n   here — use it.\n\n6. **Plan the cut list.** Decide the segments: which source, in\u002Fout times, order. Then snap to\n   the relevant grid — beats for a music montage, scene boundaries for a supercut, speech-segment\n   edges for talking-head. Respect the target length and pacing (more, shorter cuts = punchier).\n   Write it as a simple JSON cut list (see `scripts\u002Fassemble.py --help`).\n\n7. **Render.** `scripts\u002Fassemble.py` takes the cut list + options (orientation, fps, crossfade,\n   audio mode, music + duck level, captions file) and produces the final video. It re-encodes\n   uniformly, resolves paths correctly, reframes, and loudness-normalizes. Don't reimplement this.\n\n8. **Verify — always.** Probe the output (right duration \u002F aspect?), pull a few frames and look\n   at them, and sanity-check the audio (does it have the track? is the speech intact?). If\n   anything is off, diagnose and re-render. Never tell the user it's done without having looked\n   at the result yourself — a render that \"succeeded\" in ffmpeg can still be black frames,\n   wrong aspect, or silent.\n\n## Recipes\n\n`references\u002Frecipes.md` has step-by-step pipelines for the common request shapes — read it when\na brief matches one, and adapt rather than follow blindly:\n\n- **Highlight reel** — best moments from one or more clips, cut together.\n- **Beat-synced montage** — clips cut on the beat of a supplied track.\n- **Supercut** — transcribe, find every mention of a word\u002Ftopic, concat the hits.\n- **Silence removal \u002F jump-cut** — detect and drop dead air or pauses.\n- **Vertical social reel** — reframe to 9:16, best moment, burned captions, music.\n- **Soundbite \u002F talking-head** — transcribe, find the strong quotes, cut clean in\u002Fout on them.\n\n`references\u002Fffmpeg.md` is a cheat-sheet of the underlying ffmpeg filters (crop\u002Fscale, speed,\ncrossfade, overlay, mix\u002Fduck, loudnorm, drawtext\u002Fsubtitles) for when you need an operation no\nscript covers — compose a new step rather than forcing the brief into an existing one.\n\n## The toolkit\n\nAll scripts live in `scripts\u002F` and print JSON or write files; run them with the skill's venv\npython (see Setup). Each has `--help`.\n\n| Script | Does |\n|--------|------|\n| `probe.py FILE...` | media info (duration, fps, w\u002Fh, audio) as JSON — run first |\n| `proxy.py FILE -o OUT` | fast 720p proxy for analysis |\n| `analyze.py FILE [--scenes --motion --beats --silence]` | per-second interest score + peaks + beats + scene cuts + silences; caches to a sidecar |\n| `frames.py FILE --at 12,40,75 -o DIR` (or `--from-analysis`) | extract frames \u002F a contact sheet at timestamps so you can *look* and judge |\n| `transcribe.py FILE` | Whisper transcript with word\u002Fsegment timestamps (for supercuts, captions, soundbites) |\n| `assemble.py --cuts cuts.json -o OUT [--orientation --fps --music --source-audio-db --mute-source --captions ...]` | render the final cut: uniform re-encode, reframe, robust concat, music mix\u002Fduck, burned captions, loudnorm (crossfades\u002Fspeed\u002Foverlays → ffmpeg.md) |\n\n## Setup\n\nNeeds `ffmpeg`\u002F`ffprobe` on PATH. The analysis\u002Ftranscription scripts need a few Python libs;\ncreate the skill's venv once and reuse it:\n\n```bash\ncd \u003Cthis skill dir>\npython3 -m venv .venv && .venv\u002Fbin\u002Fpip install --upgrade pip\n.venv\u002Fbin\u002Fpip install -r requirements.txt        # numpy scipy librosa soundfile opencv-python scenedetect\n# whisper (only if a brief needs transcription): .venv\u002Fbin\u002Fpip install openai-whisper\n```\n\nRun scripts as `.venv\u002Fbin\u002Fpython scripts\u002F\u003Cname>.py ...`. `probe.py`, `proxy.py`, and\n`assemble.py` are pure-ffmpeg and work without the venv if you only need cutting\u002Freframing.\n\n## Gotchas (learned the hard way)\n\n- **Verify by looking, not by exit code.** ffmpeg returns 0 on plenty of broken outputs. Pull\n  frames. This is the single most important habit.\n- **Concat needs absolute paths.** The concat demuxer resolves relative entries against the\n  *list file's* directory, which silently doubles a relative workdir. `assemble.py` handles this;\n  if you hand-roll concat, write absolute paths.\n- **Re-encode to a uniform format before concatenating** clips of different size\u002Ffps\u002Fcodec, or\n  the join glitches. `assemble.py` does this.\n- **Analyze on proxies, render from originals.** Decode + optical flow + scene detection on full\n  resolution is the main time sink; a 720p proxy is ~identical for *where* the moments are.\n- **Loudness-normalize once, at the end** (`loudnorm` to ~-14 LUFS for social), after any mixing\n  — not per-clip.\n- **On macOS, media tools may need the sandbox disabled** to read files; if a probe\u002Fffmpeg call\n  is killed with no output, that's why.\n- **Cache is your friend.** `analyze.py` writes a sidecar; tweaking the edit (different length,\n  reorder, reframe) should never re-run analysis.\n","# clipsmith\n\nYou are an automatic video editor. The user gives you footage and a rough brief; you compose\nand run a **custom pipeline** for that specific request, then verify the result actually matches\nwhat they asked for. There is no fixed template — every brief implies a slightly different graph\nof operations. Your job is to pick the right operations and wire them together.\n\nThe toolkit is general (ffmpeg + a few analysis scripts), so resist the urge to hand-write\nfragile one-off ffmpeg incantations for the fiddly parts — the bundled scripts already handle\nthe things that are easy to get subtly wrong (concat path resolution, uniform re-encoding,\nreframing, loudness). Spend your effort on the *editorial* decisions instead.\n\n## The loop\n\nWork through these phases. Skip what a brief doesn't need (e.g. a \"remove the silences\" job\nneeds no vision step; a \"best moments\" job leans on it heavily), but never skip **probe** or\n**verify**.\n\n1. **Read the brief → a target.** Pin down: output length, aspect ratio (vertical 9:16 \u002F\n   square \u002F landscape), whether there's music, captions yes\u002Fno, and the *vibe* (punchy and\n   fast vs. calm and long). If two or three of these are genuinely ambiguous and would change\n   the result a lot, ask one tight question; otherwise pick sensible defaults and say what you\n   chose. Don't interrogate.\n\n2. **Probe the media.** `scripts\u002Fprobe.py FILE...` → JSON of duration, fps, resolution, audio.\n   You need this before planning anything (a 9:16 source needs no cropping; a silent clip can't\n   be silence-trimmed; a 4-minute source needs a proxy).\n\n3. **Proxy if heavy.** For anything long or high-res, make a 720p proxy first and do all\n   *analysis* on it (`scripts\u002Fproxy.py`). Render the final cut from the originals. Analysis on\n   full 4K is slow and pointless.\n\n4. **Analyze.** `scripts\u002Fanalyze.py FILE` computes, per source, a per-second \"interestingness\"\n   score (audio energy + onset + brightness, motion, scene-cut bonus) plus the music beat grid,\n   scene-cut times, and silence spans. It caches to a sidecar `.clipsmith.json`, so re-running\n   is free. Pick the signals the brief needs via flags. This finds *activity*.\n\n5. **Judge with your own eyes (the taste step).** Activity ≠ interesting. This is what separates\n   a good edit from a motion-energy slideshow. Pull candidate frames at the top peaks\n   (`scripts\u002Fframes.py FILE --at \u003Ctimes>` or `--from-analysis`) into a contact sheet, **look at\n   them**, and rank which moments actually fit the brief (\"overtake\", \"the trick lands\", \"she\n   laughs\", \"clean wide shot\"). Drop the ones that scored high on noise but look boring. For\n   longer candidates, sample 2–3 frames across the span. Your visual judgment is the feature\n   here — use it.\n\n6. **Plan the cut list.** Decide the segments: which source, in\u002Fout times, order. Then snap to\n   the relevant grid — beats for a music montage, scene boundaries for a supercut, speech-segment\n   edges for talking-head. Respect the target length and pacing (more, shorter cuts = punchier).\n   Write it as a simple JSON cut list (see `scripts\u002Fassemble.py --help`).\n\n7. **Render.** `scripts\u002Fassemble.py` takes the cut list + options (orientation, fps, crossfade,\n   audio mode, music + duck level, captions file) and produces the final video. It re-encodes\n   uniformly, resolves paths correctly, reframes, and loudness-normalizes. Don't reimplement this.\n\n8. **Verify — always.** Probe the output (right duration \u002F aspect?), pull a few frames and look\n   at them, and sanity-check the audio (does it have the track? is the speech intact?). If\n   anything is off, diagnose and re-render. Never tell the user it's done without having looked\n   at the result yourself — a render that \"succeeded\" in ffmpeg can still be black frames,\n   wrong aspect, or silent.\n\n## Recipes\n\n`references\u002Frecipes.md` has step-by-step pipelines for the common request shapes — read it when\na brief matches one, and adapt rather than follow blindly:\n\n- **Highlight reel** — best moments from one or more clips, cut together.\n- **Beat-synced montage** — clips cut on the beat of a supplied track.\n- **Supercut** — transcribe, find every mention of a word\u002Ftopic, concat the hits.\n- **Silence removal \u002F jump-cut** — detect and drop dead air or pauses.\n- **Vertical social reel** — reframe to 9:16, best moment, burned captions, music.\n- **Soundbite \u002F talking-head** — transcribe, find the strong quotes, cut clean in\u002Fout on them.\n\n`references\u002Fffmpeg.md` is a cheat-sheet of the underlying ffmpeg filters (crop\u002Fscale, speed,\ncrossfade, overlay, mix\u002Fduck, loudnorm, drawtext\u002Fsubtitles) for when you need an operation no\nscript covers — compose a new step rather than forcing the brief into an existing one.\n\n## The toolkit\n\nAll scripts live in `scripts\u002F` and print JSON or write files; run them with the skill's venv\npython (see Setup). Each has `--help`.\n\n| Script | Does |\n|--------|------|\n| `probe.py FILE...` | media info (duration, fps, w\u002Fh, audio) as JSON — run first |\n| `proxy.py FILE -o OUT` | fast 720p proxy for analysis |\n| `analyze.py FILE [--scenes --motion --beats --silence]` | per-second interest score + peaks + beats + scene cuts + silences; caches to a sidecar |\n| `frames.py FILE --at 12,40,75 -o DIR` (or `--from-analysis`) | extract frames \u002F a contact sheet at timestamps so you can *look* and judge |\n| `transcribe.py FILE` | Whisper transcript with word\u002Fsegment timestamps (for supercuts, captions, soundbites) |\n| `assemble.py --cuts cuts.json -o OUT [--orientation --fps --music --source-audio-db --mute-source --captions ...]` | render the final cut: uniform re-encode, reframe, robust concat, music mix\u002Fduck, burned captions, loudnorm (crossfades\u002Fspeed\u002Foverlays → ffmpeg.md) |\n\n## Setup\n\nNeeds `ffmpeg`\u002F`ffprobe` on PATH. The analysis\u002Ftranscription scripts need a few Python libs;\ncreate the skill's venv once and reuse it:\n\n```bash\ncd \u003Cthis skill dir>\npython3 -m venv .venv && .venv\u002Fbin\u002Fpip install --upgrade pip\n.venv\u002Fbin\u002Fpip install -r requirements.txt        # numpy scipy librosa soundfile opencv-python scenedetect\n# whisper (only if a brief needs transcription): .venv\u002Fbin\u002Fpip install openai-whisper\n```\n\nRun scripts as `.venv\u002Fbin\u002Fpython scripts\u002F\u003Cname>.py ...`. `probe.py`, `proxy.py`, and\n`assemble.py` are pure-ffmpeg and work without the venv if you only need cutting\u002Freframing.\n\n## Gotchas (learned the hard way)\n\n- **Verify by looking, not by exit code.** ffmpeg returns 0 on plenty of broken outputs. Pull\n  frames. This is the single most important habit.\n- **Concat needs absolute paths.** The concat demuxer resolves relative entries against the\n  *list file's* directory, which silently doubles a relative workdir. `assemble.py` handles this;\n  if you hand-roll concat, write absolute paths.\n- **Re-encode to a uniform format before concatenating** clips of different size\u002Ffps\u002Fcodec, or\n  the join glitches. `assemble.py` does this.\n- **Analyze on proxies, render from originals.** Decode + optical flow + scene detection on full\n  resolution is the main time sink; a 720p proxy is ~identical for *where* the moments are.\n- **Loudness-normalize once, at the end** (`loudnorm` to ~-14 LUFS for social), after any mixing\n  — not per-clip.\n- **On macOS, media tools may need the sandbox disabled** to read files; if a probe\u002Fffmpeg call\n  is killed with no output, that's why.\n- **Cache is your friend.** `analyze.py` writes a sidecar; tweaking the edit (different length,\n  reorder, reframe) should never re-run analysis.\n",{"name":5,"description":7},"main","6ef866230c1ba3459fd08fc65141dfd37fffd41a",0,"https:\u002F\u002Fgithub.com\u002Fcaezium\u002Fskills","https:\u002F\u002Fskills.henryzh.dev","https:\u002F\u002Fgithub.com\u002Fcaezium","henryzh.dev","quant, data science, full-stack, robotics.\r\nbuilding cool stuff","2026-08-24 19:21:44",{"source":8,"skillName":5,"commands":58,"prompt":68,"zipAvailable":69},[59,62,65],{"tool":60,"command":61},"npx","npx skills add caezium\u002Fskills --skill clipsmith",{"tool":63,"command":64},"bunx","bunx skills add caezium\u002Fskills --skill clipsmith",{"tool":66,"command":67},"pnpm","pnpm dlx skills add caezium\u002Fskills --skill clipsmith","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fcaezium\u002Fskills\u002Fblob\u002Fmain\u002Fclipsmith\u002FSKILL.md",true,[71,77,81,85,91,95,99,103,107],{"path":72,"resourceType":73,"resourceTypeDesc":74,"sizeBytes":75,"sourceHtmlUrl":76},"references\u002Fffmpeg.md","REFERENCE","参考文档",2853,"https:\u002F\u002Fgithub.com\u002Fcaezium\u002Fskills\u002Fblob\u002Fmain\u002Fclipsmith\u002Freferences\u002Fffmpeg.md",{"path":78,"resourceType":73,"resourceTypeDesc":74,"sizeBytes":79,"sourceHtmlUrl":80},"references\u002Frecipes.md",3596,"https:\u002F\u002Fgithub.com\u002Fcaezium\u002Fskills\u002Fblob\u002Fmain\u002Fclipsmith\u002Freferences\u002Frecipes.md",{"path":82,"resourceType":73,"resourceTypeDesc":74,"sizeBytes":83,"sourceHtmlUrl":84},"requirements.txt",63,"https:\u002F\u002Fgithub.com\u002Fcaezium\u002Fskills\u002Fblob\u002Fmain\u002Fclipsmith\u002Frequirements.txt",{"path":86,"resourceType":87,"resourceTypeDesc":88,"sizeBytes":89,"sourceHtmlUrl":90},"scripts\u002Fanalyze.py","SCRIPT","脚本",8240,"https:\u002F\u002Fgithub.com\u002Fcaezium\u002Fskills\u002Fblob\u002Fmain\u002Fclipsmith\u002Fscripts\u002Fanalyze.py",{"path":92,"resourceType":87,"resourceTypeDesc":88,"sizeBytes":93,"sourceHtmlUrl":94},"scripts\u002Fassemble.py",6452,"https:\u002F\u002Fgithub.com\u002Fcaezium\u002Fskills\u002Fblob\u002Fmain\u002Fclipsmith\u002Fscripts\u002Fassemble.py",{"path":96,"resourceType":87,"resourceTypeDesc":88,"sizeBytes":97,"sourceHtmlUrl":98},"scripts\u002Fframes.py",3719,"https:\u002F\u002Fgithub.com\u002Fcaezium\u002Fskills\u002Fblob\u002Fmain\u002Fclipsmith\u002Fscripts\u002Fframes.py",{"path":100,"resourceType":87,"resourceTypeDesc":88,"sizeBytes":101,"sourceHtmlUrl":102},"scripts\u002Fprobe.py",2250,"https:\u002F\u002Fgithub.com\u002Fcaezium\u002Fskills\u002Fblob\u002Fmain\u002Fclipsmith\u002Fscripts\u002Fprobe.py",{"path":104,"resourceType":87,"resourceTypeDesc":88,"sizeBytes":105,"sourceHtmlUrl":106},"scripts\u002Fproxy.py",1092,"https:\u002F\u002Fgithub.com\u002Fcaezium\u002Fskills\u002Fblob\u002Fmain\u002Fclipsmith\u002Fscripts\u002Fproxy.py",{"path":108,"resourceType":87,"resourceTypeDesc":88,"sizeBytes":109,"sourceHtmlUrl":110},"scripts\u002Ftranscribe.py",1962,"https:\u002F\u002Fgithub.com\u002Fcaezium\u002Fskills\u002Fblob\u002Fmain\u002Fclipsmith\u002Fscripts\u002Ftranscribe.py",[112,133,161,179,200,222],{"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":126,"publishedAt":43},"01M71JT4RYCAS18RF6003YYR1D","writing-for-agents","Writing documents for agents. Use when creating or editing skills, or modifying AGENTS.md or CLAUDE.md.","https:\u002F\u002Fgithub.com\u002Fcaezium\u002Fskills\u002Fblob\u002Fmain\u002Fwriting-for-agents\u002FSKILL.md",[118,121],{"categoryNo":119,"categoryName":120,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000018","写作与文案",{"categoryNo":122,"categoryName":123,"taxonomyType":124,"taxonomyTypeDesc":125},"01JBSEEDCAT000000000000036","内容创作者","PROFESSION","适用职业",[127,128,129,130,131,132,28],"writing for agents","for","writing","editing","use","documents",{"skillNo":134,"name":135,"nameZh":6,"description":136,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":137,"categories":138,"keywords":142,"publishedAt":43},"01M71JT0GES4AZ5ADH8EADNN1Z","wrangler","Cloudflare Workers CLI for deploying, developing, and managing Workers, KV, R2, D1, Vectorize, Hyperdrive, Workers AI, Containers, Queues, Workflows, Pipelines, and Secrets Store. Load before running wrangler commands to ensure correct syntax and best practices. Biases towards retrieval from Cloudflare docs over pre-trained knowledge.","https:\u002F\u002Fgithub.com\u002Fcaezium\u002Fskills\u002Fblob\u002Fmain\u002Fwrangler\u002FSKILL.md",[139],{"categoryNo":140,"categoryName":141,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000010","后端与接口",[135,143,144,145,146,147,148,149,150,151,152,153,154,128,155,156,157,158,159,160],"cloudflare","d1","knowledge","workflows","cli","secrets","commands","syntax","cloudflare workers","r2","queues","running","containers","correct","load","workers","best practices","store",{"skillNo":162,"name":163,"nameZh":6,"description":164,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":165,"categories":166,"keywords":167,"publishedAt":43},"01M71JSWX4R203M32K9A18AAW8","workers-best-practices","Reviews and authors Cloudflare Workers code against production best practices. Load when writing new Workers, reviewing Worker code, configuring wrangler.jsonc, or checking for common Workers anti-patterns (streaming, floating promises, global state, secrets, bindings, observability). Biases towards retrieval from Cloudflare docs over pre-trained knowledge.","https:\u002F\u002Fgithub.com\u002Fcaezium\u002Fskills\u002Fblob\u002Fmain\u002Fworkers-best-practices\u002FSKILL.md",[],[168,158,159,169,170,171,128,143,172,145,173,135,129,157,28,174,175,176,177,178],"workers best practices","observability","reviews","promises","worker","new","code","common","streaming","state","anti-patterns",{"skillNo":180,"name":181,"nameZh":6,"description":182,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":183,"categories":184,"keywords":185,"publishedAt":43},"01M71JST6P88MXGT2MWDTPJGEG","wizard","Generate an interactive bash wizard that walks a human through steps only they can perform. Use when provisioning infrastructure, setting up credentials or CI secrets, walking an unfamiliar third-party dashboard, or running a one-off migration or cutover. Don't invoke this for steps the agent can perform itself.","https:\u002F\u002Fgithub.com\u002Fcaezium\u002Fskills\u002Fblob\u002Fmain\u002Fwizard\u002FSKILL.md",[],[181,154,186,128,187,188,189,190,131,191,28,192,193,194,195,196,197,198,199,148],"setting","up","ci","provisioning infrastructure","only","invoke","Bash","one","interactive","can","credentials","dashboard","generate","human",{"skillNo":201,"name":202,"nameZh":6,"description":203,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":204,"sourceHtmlUrl":205,"categories":206,"keywords":207,"publishedAt":43},"01M71JSQ95S7BQ61DAPN95X2W4","why","Use for 'why does X work this way', 'why we picked Y', design rationale, regressions, postmortems, or data-backed thresholds. Discovers available evidence sources, searches each relevant category, and returns a cited read on decisions and tradeoffs. Use how for runtime behavior.","https:\u002F\u002Fgithub.com\u002Fcaezium\u002Fskills\u002Fblob\u002Fmain\u002Fwhy\u002FLICENSE","https:\u002F\u002Fgithub.com\u002Fcaezium\u002Fskills\u002Fblob\u002Fmain\u002Fwhy\u002FSKILL.md",[],[202,208,209,210,211,128,212,213,131,214,215,216,217,218,219,220,221],"sources","thresholds","evidence","relevant","runtime","design","decisions","work","how","behavior","read","data","rationale","returns",{"skillNo":223,"name":224,"nameZh":6,"description":225,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":226,"categories":227,"keywords":231,"publishedAt":43},"01M71JSMHFVSH8HK8GXN6YAX6J","week-prep","Generate a briefing with calendar, tasks, and attention flags. Use for \"week prep\", \"briefing\", or \"what's on this week\".","https:\u002F\u002Fgithub.com\u002Fcaezium\u002Fskills\u002Fblob\u002Fmain\u002Fweek-prep\u002FSKILL.md",[228],{"categoryNo":229,"categoryName":230,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000023","Skill与智能体开发",[232,233,234,235,236,128,237,238,239,240,131,192,241,242,243,198,244,218,245,246],"week prep","prep","deep","attention","get","search","messages","briefing","MCP","tasks","with","slack","calendar","flags","grep"]