[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M74XMF8WM3RRP8JF37RT6R1P":3,"skill-same:01M74XMF8WM3RRP8JF37RT6R1P":53},{"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":23,"commitSha":24,"skillDirectory":25,"forksCount":13,"repositoryHtmlUrl":26,"repositoryHomepageUrl":6,"ownerHtmlUrl":27,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":28,"install":29,"resources":42},"01M74XMF8WM3RRP8JF37RT6R1P","scan-capture-quality-gates",null,"Two scan-capture quality failures — (1) blurry captures sail through with no warning, OR the opposite, a hard blur gate rejects every low-texture subject (plain box, matte fabric, blank wall) and re-prompts \"Retake\" forever, stranding the user mid-scan; (2) an item auto-names itself with background OCR prose like \"Complete the form below and\" and the UI renders that guess as high-confidence (green). Use when adding a sharpness\u002Fblur accept-reject gate to a camera capture pipeline, or when deriving item\u002Fdocument names from OCR text in a scanning app.","abhijitbansal\u002Fclaude-skills","abhijitbansal","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F271802621?v=4","01M74XA5M33XYNJ2D2TJDS7W9E","01M74XA5MQ8GMWJW10ZJ4V0NVG",1,"MIT","https:\u002F\u002Fgithub.com\u002Fabhijitbansal\u002Fclaude-skills\u002Fblob\u002Fmain\u002Fplugins\u002Fios-dev\u002Fskills\u002Fscan-capture-quality-gates\u002FSKILL.md",[],[],"2026-11-12 13:55:43",false,"---\nname: scan-capture-quality-gates\ndescription: Two scan-capture quality failures — (1) blurry captures sail through with no warning, OR the opposite, a hard blur gate rejects every low-texture subject (plain box, matte fabric, blank wall) and re-prompts \"Retake\" forever, stranding the user mid-scan; (2) an item auto-names itself with background OCR prose like \"Complete the form below and\" and the UI renders that guess as high-confidence (green). Use when adding a sharpness\u002Fblur accept-reject gate to a camera capture pipeline, or when deriving item\u002Fdocument names from OCR text in a scanning app.\n---\n\n# Scan Capture Quality Gates: Soft Blur Gate + Auto-Naming Discipline\n\n## Symptom\n\n- **Blur:** captures of low-texture subjects (plain cardboard, matte fabric)\n  fail a hard sharpness check every single attempt — the user is stuck in a\n  retake loop and can never finish the scan. Or, with no gate at all, blurry\n  frames are silently accepted and OCR\u002Frecognition downstream degrades.\n- **Naming:** a freshly scanned item appears named with a fragment of\n  background OCR prose — cubby's verbatim case: **\"Complete the form below\n  and\"** — and the name renders in the high-confidence (green) style, so the\n  user trusts it.\n\n## Root cause\n\n- Sharpness (variance of Laplacian) is a **texture** metric, not a focus\n  metric. Low-texture subjects score low even in perfect focus, so any *hard*\n  threshold either passes blur (too low) or strands users (too high). A\n  threshold of 80 rejects legitimate subjects; ~50 is the working value.\n- Auto-naming leaks when there are **multiple creation entry points**: any\n  caller that can pass a name string will eventually pass raw OCR output, and\n  OCR confidence scores are per-glyph, not \"is this a *name*\" — prose reads as\n  high-confidence text.\n\n## Fix 1 — variance-of-Laplacian as a SOFT gate\n\nThree verdicts, not pass\u002Ffail. Auto-accept after a retry budget so low-texture\nsubjects can't strand the user.\n\n```swift\nnonisolated enum ScanConstants {\n    static let sharpnessThreshold: Double = 50  \u002F\u002F NOT 80 — 80 strands low-texture subjects\n    static let maxBlurRetries = 2               \u002F\u002F then auto-accept\n}\n\nnonisolated enum BlurVerdict {\n    case deliver            \u002F\u002F sharp enough — accept silently\n    case warnAllowOverride  \u002F\u002F soft warn: offer Retake, but keep a \"Use Anyway\" path\n    case autoAccept         \u002F\u002F retry budget exhausted — accept; optionally tag low-detail\n}\n\nnonisolated enum BlurGate {\n    static func verdict(sharpness: Double, retryCount: Int) -> BlurVerdict {\n        if sharpness >= ScanConstants.sharpnessThreshold { return .deliver }\n        if retryCount >= ScanConstants.maxBlurRetries { return .autoAccept }\n        return .warnAllowOverride\n    }\n}\n```\n\nSharpness itself is measured as the variance of the Laplacian over an 8-bit\nluma plane (higher = sharper) — a vImage convolution, not a decision rule, so\nit lives out-of-line.\n\n**Read `references\u002Fblur-gate.md` before implementing** — `laplacianVariance`,\nthe vImage\u002FAccelerate sharpness measurement that feeds `verdict` above.\n\nRules: the gate is advisory (`warnAllowOverride` always keeps a \"Use Anyway\"\nbutton); never block delivery of the frame itself — gate the *accept* step.\n\n## Fix 2 — auto-naming discipline\n\nONE creation entry point (callers cannot inject a name), a confidence floor, a\nsentence-shape reject list, and a hard cap on OCR-derived confidence so OCR\nguesses never render as verified (green) in the UI.\n\n```swift\nnonisolated enum NamingConstants {\n    static let minNameConfidence: Double = 0.5   \u002F\u002F below the floor → default name\n    static let ocrNameConfidenceCap: Double = 0.69 \u002F\u002F OCR names never reach \"green\"\n}\n\nnonisolated struct NameCandidate { let text: String; let confidence: Double }\n```\n\n`ItemNamer.resolveName` is the ONLY way an item gets a name (no caller can\ninject a name string): it filters out sentence-like candidates, caps every\nsurviving candidate's confidence at `ocrNameConfidenceCap`, takes the best,\nand falls back to `defaultName` if nothing clears `minNameConfidence`.\n`isSentenceLike` is the prose-fragment reject rule — text is sentence-like\n(not a name) if it has more than 6 words, ends in `.`\u002F`,`\u002F`:`, or contains\n2+ words from a prose-marker set (the\u002Fand\u002Fyour\u002Fplease\u002Fbelow\u002Fabove\u002Fcomplete\u002F\nenter\u002Ffill). This is what stops fragments like \"Complete the form below and\"\nfrom becoming an item name.\n\n**Read `references\u002Focr-name-scoring.md` before implementing** — the full\n`ItemNamer.resolveName` and `isSentenceLike` implementations.\n\n## Evidence\n\n- **cubby** — P4\u002FP6 fix series: soft blur gate (`.deliver` \u002F\n  `.warnAllowOverride` \u002F `.autoAccept`, threshold lowered 80 → 50, auto-accept\n  after retry budget) and the auto-naming cleanup after background OCR prose\n  (\"Complete the form below and\") became an item name.\n- **a private document-scanner app** — shares the auto-naming discipline; had *no*\n  sharpness gate at all (the gap this skill closes for new scanning apps).\n- Mining report Themes 3.3–3.4 (`docs\u002Fsuperpowers\u002Fspecs\u002F2026-07-01-ios-mining-report.md`).\n\n## Related skills\n\n- `avfoundation-capture-delivery-watchdog` — frames never *arriving* (delivery\n  stalls) is a different failure than frames arriving blurry; gate quality here,\n  watchdog delivery there.\n- `vision-barcode-cidetector-fallback` — recognition fallbacks on the captured\n  frame; a soft blur gate reduces how often those fallbacks fire.\n- `swift6-mainactor-compile-fixes` — `BlurGate`\u002F`ItemNamer` are pure compute:\n  keep them `nonisolated` so they run off-main in the capture pipeline.\n- `async-enrichment-silent-loss-outcome-states` — downstream on the same scan\n  pipeline: once a frame passes this gate, its async AI enrichment result\n  still needs a visible outcome state, win or lose the merge.\n","# Scan Capture Quality Gates: Soft Blur Gate + Auto-Naming Discipline\n\n## Symptom\n\n- **Blur:** captures of low-texture subjects (plain cardboard, matte fabric)\n  fail a hard sharpness check every single attempt — the user is stuck in a\n  retake loop and can never finish the scan. Or, with no gate at all, blurry\n  frames are silently accepted and OCR\u002Frecognition downstream degrades.\n- **Naming:** a freshly scanned item appears named with a fragment of\n  background OCR prose — cubby's verbatim case: **\"Complete the form below\n  and\"** — and the name renders in the high-confidence (green) style, so the\n  user trusts it.\n\n## Root cause\n\n- Sharpness (variance of Laplacian) is a **texture** metric, not a focus\n  metric. Low-texture subjects score low even in perfect focus, so any *hard*\n  threshold either passes blur (too low) or strands users (too high). A\n  threshold of 80 rejects legitimate subjects; ~50 is the working value.\n- Auto-naming leaks when there are **multiple creation entry points**: any\n  caller that can pass a name string will eventually pass raw OCR output, and\n  OCR confidence scores are per-glyph, not \"is this a *name*\" — prose reads as\n  high-confidence text.\n\n## Fix 1 — variance-of-Laplacian as a SOFT gate\n\nThree verdicts, not pass\u002Ffail. Auto-accept after a retry budget so low-texture\nsubjects can't strand the user.\n\n```swift\nnonisolated enum ScanConstants {\n    static let sharpnessThreshold: Double = 50  \u002F\u002F NOT 80 — 80 strands low-texture subjects\n    static let maxBlurRetries = 2               \u002F\u002F then auto-accept\n}\n\nnonisolated enum BlurVerdict {\n    case deliver            \u002F\u002F sharp enough — accept silently\n    case warnAllowOverride  \u002F\u002F soft warn: offer Retake, but keep a \"Use Anyway\" path\n    case autoAccept         \u002F\u002F retry budget exhausted — accept; optionally tag low-detail\n}\n\nnonisolated enum BlurGate {\n    static func verdict(sharpness: Double, retryCount: Int) -> BlurVerdict {\n        if sharpness >= ScanConstants.sharpnessThreshold { return .deliver }\n        if retryCount >= ScanConstants.maxBlurRetries { return .autoAccept }\n        return .warnAllowOverride\n    }\n}\n```\n\nSharpness itself is measured as the variance of the Laplacian over an 8-bit\nluma plane (higher = sharper) — a vImage convolution, not a decision rule, so\nit lives out-of-line.\n\n**Read `references\u002Fblur-gate.md` before implementing** — `laplacianVariance`,\nthe vImage\u002FAccelerate sharpness measurement that feeds `verdict` above.\n\nRules: the gate is advisory (`warnAllowOverride` always keeps a \"Use Anyway\"\nbutton); never block delivery of the frame itself — gate the *accept* step.\n\n## Fix 2 — auto-naming discipline\n\nONE creation entry point (callers cannot inject a name), a confidence floor, a\nsentence-shape reject list, and a hard cap on OCR-derived confidence so OCR\nguesses never render as verified (green) in the UI.\n\n```swift\nnonisolated enum NamingConstants {\n    static let minNameConfidence: Double = 0.5   \u002F\u002F below the floor → default name\n    static let ocrNameConfidenceCap: Double = 0.69 \u002F\u002F OCR names never reach \"green\"\n}\n\nnonisolated struct NameCandidate { let text: String; let confidence: Double }\n```\n\n`ItemNamer.resolveName` is the ONLY way an item gets a name (no caller can\ninject a name string): it filters out sentence-like candidates, caps every\nsurviving candidate's confidence at `ocrNameConfidenceCap`, takes the best,\nand falls back to `defaultName` if nothing clears `minNameConfidence`.\n`isSentenceLike` is the prose-fragment reject rule — text is sentence-like\n(not a name) if it has more than 6 words, ends in `.`\u002F`,`\u002F`:`, or contains\n2+ words from a prose-marker set (the\u002Fand\u002Fyour\u002Fplease\u002Fbelow\u002Fabove\u002Fcomplete\u002F\nenter\u002Ffill). This is what stops fragments like \"Complete the form below and\"\nfrom becoming an item name.\n\n**Read `references\u002Focr-name-scoring.md` before implementing** — the full\n`ItemNamer.resolveName` and `isSentenceLike` implementations.\n\n## Evidence\n\n- **cubby** — P4\u002FP6 fix series: soft blur gate (`.deliver` \u002F\n  `.warnAllowOverride` \u002F `.autoAccept`, threshold lowered 80 → 50, auto-accept\n  after retry budget) and the auto-naming cleanup after background OCR prose\n  (\"Complete the form below and\") became an item name.\n- **a private document-scanner app** — shares the auto-naming discipline; had *no*\n  sharpness gate at all (the gap this skill closes for new scanning apps).\n- Mining report Themes 3.3–3.4 (`docs\u002Fsuperpowers\u002Fspecs\u002F2026-07-01-ios-mining-report.md`).\n\n## Related skills\n\n- `avfoundation-capture-delivery-watchdog` — frames never *arriving* (delivery\n  stalls) is a different failure than frames arriving blurry; gate quality here,\n  watchdog delivery there.\n- `vision-barcode-cidetector-fallback` — recognition fallbacks on the captured\n  frame; a soft blur gate reduces how often those fallbacks fire.\n- `swift6-mainactor-compile-fixes` — `BlurGate`\u002F`ItemNamer` are pure compute:\n  keep them `nonisolated` so they run off-main in the capture pipeline.\n- `async-enrichment-silent-loss-outcome-states` — downstream on the same scan\n  pipeline: once a frame passes this gate, its async AI enrichment result\n  still needs a visible outcome state, win or lose the merge.\n",{"name":5,"description":7},"main","153711dd9432743d97f0b33b080ada5798ea623c","plugins\u002Fios-dev\u002Fskills\u002Fscan-capture-quality-gates","https:\u002F\u002Fgithub.com\u002Fabhijitbansal\u002Fclaude-skills","https:\u002F\u002Fgithub.com\u002Fabhijitbansal","2026-11-07 12:59:51",{"source":8,"skillName":5,"commands":30,"prompt":40,"zipAvailable":41},[31,34,37],{"tool":32,"command":33},"npx","npx skills add abhijitbansal\u002Fclaude-skills --skill scan-capture-quality-gates",{"tool":35,"command":36},"bunx","bunx skills add abhijitbansal\u002Fclaude-skills --skill scan-capture-quality-gates",{"tool":38,"command":39},"pnpm","pnpm dlx skills add abhijitbansal\u002Fclaude-skills --skill scan-capture-quality-gates","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fabhijitbansal\u002Fclaude-skills\u002Fblob\u002Fmain\u002Fplugins\u002Fios-dev\u002Fskills\u002Fscan-capture-quality-gates\u002FSKILL.md",true,[43,49],{"path":44,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":47,"sourceHtmlUrl":48},"references\u002Fblur-gate.md","REFERENCE","参考文档",1260,"https:\u002F\u002Fgithub.com\u002Fabhijitbansal\u002Fclaude-skills\u002Fblob\u002Fmain\u002Fplugins\u002Fios-dev\u002Fskills\u002Fscan-capture-quality-gates\u002Freferences\u002Fblur-gate.md",{"path":50,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":51,"sourceHtmlUrl":52},"references\u002Focr-name-scoring.md",1547,"https:\u002F\u002Fgithub.com\u002Fabhijitbansal\u002Fclaude-skills\u002Fblob\u002Fmain\u002Fplugins\u002Fios-dev\u002Fskills\u002Fscan-capture-quality-gates\u002Freferences\u002Focr-name-scoring.md",[54,61,68,75,82,89],{"skillNo":55,"name":56,"nameZh":6,"description":57,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":58,"categories":59,"keywords":60,"publishedAt":18},"01M74XTV2V7GZM8FDBXQZZ74VX","second-wind","Orchestrate long unattended Claude Code runs across multiple repos with the wind CLI. Use when the user wants to run Claude Code overnight, resume sessions after the 5-hour usage limit, run Claude in many repos at once via tmux, or says \"set up second wind\", \"wind up\", \"overnight run\", \"resume after limit\", \"usage limit orchestrator\".","https:\u002F\u002Fgithub.com\u002Fabhijitbansal\u002Fclaude-skills\u002Fblob\u002Fmain\u002Fplugins\u002Fsecond-wind\u002Fskills\u002Fsecond-wind\u002FSKILL.md",[],[],{"skillNo":62,"name":63,"nameZh":6,"description":64,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":65,"categories":66,"keywords":67,"publishedAt":18},"01M74XTPGN1HDQSKGCZZX0AZPC","structured-output-adaptive-thinking-token-bloat","A Claude API call using output_config.format (json_schema structured outputs) with thinking:{type:\"adaptive\"} burns most of its output budget on thinking (observed ~21k of ~24k tokens) without improving JSON quality, making the run up to ~5x more expensive than the same call with thinking omitted — use when calling the Messages API with a json_schema output format for extraction, ranking, classification, or summarization tasks, or when a schema-constrained call's usage.output_tokens is far larger than the returned JSON.","https:\u002F\u002Fgithub.com\u002Fabhijitbansal\u002Fclaude-skills\u002Fblob\u002Fmain\u002Fplugins\u002Fprompt-craft\u002Fskills\u002Fstructured-output-adaptive-thinking-token-bloat\u002FSKILL.md",[],[],{"skillNo":69,"name":70,"nameZh":6,"description":71,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":72,"categories":73,"keywords":74,"publishedAt":18},"01M74XTKTRHAMC2JFHQ1P1E6GR","review","Apply a code-review lens to a diff, branch, or file — find correctness bugs, security issues, and quality problems, ranked by severity, with a concrete fix for each. Use when the user says \"review this\", \"review my diff\", \"check this PR\", \"look this over before I commit\", or \"is this safe\". Skip when the user wants the code written, not reviewed.","https:\u002F\u002Fgithub.com\u002Fabhijitbansal\u002Fclaude-skills\u002Fblob\u002Fmain\u002Fplugins\u002Fprompt-craft\u002Fskills\u002Freview\u002FSKILL.md",[],[],{"skillNo":76,"name":77,"nameZh":6,"description":78,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":79,"categories":80,"keywords":81,"publishedAt":18},"01M74XTGGCW7GWDDHPEMJQC7Y1","refresh","Rebuild the prompt-craft command registry and usage profile, and print a summary of installed commands and your personalized recommendations. Use when you install or update plugins and want the advisor to pick them up immediately, or say \"refresh prompt-craft\", \"rebuild the command registry\".","https:\u002F\u002Fgithub.com\u002Fabhijitbansal\u002Fclaude-skills\u002Fblob\u002Fmain\u002Fplugins\u002Fprompt-craft\u002Fskills\u002Frefresh\u002FSKILL.md",[],[],{"skillNo":83,"name":84,"nameZh":6,"description":85,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":86,"categories":87,"keywords":88,"publishedAt":18},"01M74XTCNGNR4G19P3Z95KTHQJ","refactor","Apply a refactoring lens — restructure code without changing behavior, guarded by tests that pass before and after. Use when the user says \"refactor\", \"clean this up\", \"extract this\", \"simplify this\", \"reduce duplication\", or \"make this readable\" on existing working code. Skip when the change is meant to alter behavior (that's a feature or a fix).","https:\u002F\u002Fgithub.com\u002Fabhijitbansal\u002Fclaude-skills\u002Fblob\u002Fmain\u002Fplugins\u002Fprompt-craft\u002Fskills\u002Frefactor\u002FSKILL.md",[],[],{"skillNo":90,"name":91,"nameZh":6,"description":92,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":93,"categories":94,"keywords":95,"publishedAt":18},"01M74XT9FKXN3N106G7Q3X56XR","plan","Decompose a non-trivial task into an explicit, tracked plan with goals and per-step acceptance criteria before writing code. Use when the user says \"plan this\", \"break this down\", \"what's the plan\", or asks for a multi-step feature\u002Frefactor that touches more than one file. Pairs with plan mode. Skip for single-edit changes that need no decomposition.","https:\u002F\u002Fgithub.com\u002Fabhijitbansal\u002Fclaude-skills\u002Fblob\u002Fmain\u002Fplugins\u002Fprompt-craft\u002Fskills\u002Fplan\u002FSKILL.md",[],[]]