20:"$Sreact.view_transition" 21:I[9754,["/_next/static/chunks/2qcf_o2xotudn.js","/_next/static/chunks/1gyqeidjm5eoa.js","/_next/static/chunks/0_fjnxc4tt_l_.js","/_next/static/chunks/1ymt1shyhmu-4.js","/_next/static/chunks/3kv2zf0i7s2wc.js","/_next/static/chunks/1ucasf31l8j2v.js","/_next/static/chunks/0yj6_8vip2oqy.js","/_next/static/chunks/2l5xdnifczg1b.js","/_next/static/chunks/192kgs9v_t14j.js","/_next/static/chunks/0z6pw0v5c_lx5.js","/_next/static/chunks/08r60e6icmtkc.js","/_next/static/chunks/3trx6c0i7_x-3.js","/_next/static/chunks/40sro5bzp9bi_.js","/_next/static/chunks/1wgf3tk38y6cf.js"],"ProjectDetail"] 22:T4a21, VibeSkills · 路由枢纽

vibe-skills · 通用型 Skill 路由枢纽

一条枢纽，把 100+ 本地 Skill 收拢成每一趟明确的路由

一个通用型 Skill：自动识别任务意图，把本地 Skills 逐一调度到唯一对应的工作轨，再沿 harness 工作流把每个模块编排到交付。

Agent 编排 v4.0.0 Apache-2.0

github.com/foryourhealth111-pixel/vibe-skills

可选 Skills

100+

Hub 收到的全部本地信号

选中

本任务实际路由到的 Skills

工作单元

整理成 5 个工作组与 10 个单元

//1

信号进入端点，随即进入枢纽

VibeSkills 并不堆叠技能列表，而是以一次规划状态机把整段任务做模块化分解：接收原始意图，拆成模块，再为每个模块调度一个 Skill。枢纽只做一件事——把每一路输入送到恰好一条轨道。

接收任务

确认需求 / 规模

分解模块

组织工作组 ↗

分配工作单元 ↗

执行并记录 ↗

检查结果 ✕

//2

信号到轨道：每个 Skill 只落一条线

当一条路由信号抵达，枢纽不再让多个候选竞争，而是把它唯一地指派到一条 teal 主轨——这一条才是本模块真正要走的路线。

01路由输入（原生信号）

02枢纽评估 → 判定规模（L / XL）

03指派到唯一工作轨

04进入组 / 单元 → 交付检查

→模块执行的交接约定 module-execution.json

→v4.0.0 主执行交接版本

→经 vibe 入口重新进入枢纽

//3

一场机器学习实验的拆解

在案列场景里，枢纽只接收一次任务——随后把 100+ Skills 收拢为 7 个，组织成 5 个工作组、10 个工作单元。密度先收束，再展开成少数明确路线。

100+

候选 Skills

选中

5 组

工作组

10 单元

工作单元

SIGNAL LAYER

许多进来

100+

一条枢纽

路由一次

七条出去

7 Skills

//4

交付轨道：一次到底，17/17 全绿

任务从接收一路走到检查交付，不中断。承接上文的路由量级，这一次任务的末端是一份闭合的成绩单——逐项核对，全数通过。

L / XL规模判定

vibe枢纽重入

↩module-execution

5工作组

10工作单元

7选中 Skills

DELIVERY

17 / 17

检查全部通过

//5

接入枢纽之前，先读条件

VibeSkills 是本地安装到指定 Skills 目录的 Skill。v4.0.0 移除了旧入口——枢纽只经新的 vibe 命令进入。

安装与接入条件

→安装到指定 Skills 目录，例如 ~/.agents/skills
→v4.0.0 移除公开旧入口，请使用 vibe 命令重新进入，并返回规范的 module-execution.json
→v4.0.0 不安装或推荐 chrome、chrome-devtools、playwright、context7、claude-flow 等 MCP

23:T481a, VibeSkills · Routing Switchboard

vibe-skills · a general-purpose routing skill

A switchboard that folds 100+ local Skills into one routed track per task

A general-purpose Skill: it reads the incoming task, dispatches each local Skill to exactly one lane, then orchestrates every module along the harness workflow through to delivery.

Agent Orchestration v4.0.0 Apache-2.0

github.com/foryourhealth111-pixel/vibe-skills

Available Skills

100+

Every local signal reaching the hub

Selected

Skills actually routed for this task

Work Units

Settled into 5 groups and 10 units

//1

Signal enters a port, then the hub

VibeSkills does not stack a skill list. It runs a planning state machine that decomposes the whole task: take the raw intent, split it into modules, and dispatch a Skill to each module. The hub does one thing — sends every input down exactly one lane.

receive task

confirm needs / size

decompose modules

organize groups →

assign units →

execute and log →

check outcome ✕

//2

Signal to lane: one Skill lands on one wire

When a routing signal arrives, the hub no longer lets candidates compete. It hands the module to exactly one teal trunk lane — the one line that module is going to travel.

01routing input (raw signal)

02hub evaluates → sizes task (L / XL)

03assigned to one trunk lane

04into group / unit → delivery check

→execution handoff contract module-execution.json

→v4.0.0 major execution-handoff release

→re-enter the hub through the vibe entry

//3

One machine-learning experiment, split open

In the example, the hub receives a single task — then folds 100+ Skills into 7, organized as 5 working groups of 10 work units. Density draws in first, then fans out into a few clear routes.

100+

candidate Skills

selected

5 groups

working groups

10 units

work units

SIGNAL LAYER

Many in

100+

One hub

routed once

Seven out

7 Skills

//4

Delivery track: one clean pass, 17/17 green

The task runs from intake to checked handoff without a break. Picking up the case above, the run ends on a closed statement of work — item by item, every check passing.

L / XLsize decision

vibehub re-entry

←module-execution

5working groups

10work units

7selected Skills

DELIVERY

17 / 17

all checks passed

//5

Read the conditions before wiring in

VibeSkills installs locally into a specific Skills directory. On v4.0.0 the legacy entry point is gone — the hub is only reached through the new vibe command.

Install & access conditions

→Install into a specific Skills directory, e.g. ~/.agents/skills
→v4.0.0 removes the legacy public entry; use the vibe command to re-enter and return a canonical module-execution.json
→v4.0.0 does not install or recommend chrome, chrome-devtools, playwright, context7, or claude-flow MCPs

a:[["$","$20",null,{"enter":{"secondary-forward":"secondary-forward","secondary-back":"secondary-back-enter","default":"none"},"exit":{"secondary-forward":"secondary-forward","secondary-back":"secondary-back-exit","default":"none"},"update":{"secondary-forward":"secondary-forward","secondary-back":"secondary-back-update","default":"none"},"default":"none","children":["$","$L21","cmsndp4uo0075cz472o9by3ym:zh-CN:beautified",{"project":{"id":"cmsndp4uo0075cz472o9by3ym","slug":"foryourhealth111-pixel-vibe-skills","name":"vibe-skills","description":"通用型 Skill，自动路由并编排本地 Skills 与 harness 工作流。","authorName":"foryourhealth111-pixel","authorVerification":{"verified":false,"verifiedAt":null},"category":"development_tools","subcategory":"data_ai_tools","experienceUrl":null,"qualityScore":67,"featured":false,"officialSite":null,"views":{"original":{"available":true},"official":{"available":false,"reason":"disabled_by_policy"},"beautified":{"available":true,"reason":null},"defaultView":"beautified"},"beautifiedView":{"revisionId":"cmt2zh06z00mt6271t2q52qpt","kind":"artifact","artifact":{"schemaVersion":"localized-static-html-v4","kind":"static_html","locales":["zh-CN","en"],"defaultLocale":"zh-CN","documents":{"zh-CN":"$22","en":"$23"},"designDirections":{"schemaVersion":"dodot-design-directions-v1","directions":[{"id":"routing-switchboard","name":"Routing Switchboard","projectMetaphor":"A junction that takes many unlabeled incoming signals and dispatches each to exactly one wired lane — the essence of VibeSkills routing 100+ local Skills down to a chosen few routed modules per task.","signatureVisual":"A wire-fan junction: a dense bundle of thin signal lines (one per available local Skill) converges into a central switch hub, then fans out into fewer, thicker routed trunk lines labeled with the organized working groups. The density contrast (many-in, few-out) carries the routing claim.","heroGeometry":"A radial convergence fan: a wide arc of many fine lines drawn into a central trapezoid hub, then re-expanding into several bold routed paths with numbered pins.","sectionSequence":["convergence-hero","routing-mechanism-stage","signal-to-lane-mapping","case-breakdown-splits","delivery-track-results","install-condition"],"surfaceModel":"Signal planes: layered translucent paper-like panels that stack like routing layers, with thin connective rule-lines passing between panels as if wires thread seams. Sharp knuckle joints where lines change direction.","palette":"Ink on warm paper: near-black ink #16130F, warm cream field #F4EFE6, junction signal amber #E8A33D, routed trunk teal #1F6E63. Signals as amber threads, routed lanes as teal threads.","typography":"Monospaced display voice (thread/nodes feel) for headings paired with a humanist sans for routed body text. Numeric routing IDs rendered in tabular mono.","desktopComposition":"Left-anchored headline over the convergence fan on the right; the fan visually shrinks a field of fine lines into three routed trunk columns that carry the route→group→outcome story downward into the case section.","mobileTransformation":"The fan compresses into a vertical spine: fine input lines run down the left as a taper column, bottleneck into a single hub dot, then tilt into stacked routed lanes that read as scroll sequences rather than a side-by-side graph.","mediaStrategy":"No media region — assetManifest is empty and no local asset is safe. The signature wire-fan junction is a derived illustration of verified routing/group relationships, not a product screenshot.","forbiddenPatterns":["browser/terminal/editor shell","three-metric-card hero","radial gradient glow or particle field","fake product screenshot or placeholder frame"],"compositionFingerprint":"many-to-few convergence fan with signal-thread rule-lines threading between translucent route-layer panels"},{"id":"planning-state-machine","name":"Planning State Machine","projectMetaphor":"A harness pipeline of states — a planning state machine that decomposes a task and advances module-by-module until delivery, mirroring the Superpower/GSD-style stage flow.","signatureVisual":"A horizontal stepped track with distinct state stations (intake, level, decompose, group, execute, check) each shown as a module-shaped marker on a rail, with an advance arrow only where evidence supports a next transition.","heroGeometry":"A stepped monotone line that climbs one stage at a time: a broken x-axis path whose plateaus are the routing/grouping modules and whose final upright is the delivery check.","sectionSequence":["state-rail-hero","state-transition-model","module-decomposition-step","level-recommendation-stage","case-ward-execution","delivery-gate-checks"],"surfaceModel":"Track and rail surfaces: cards sit as discrete stations bolted to a continuous baseline rail. A visible vertical seam separates hold (pre-decompose) and run (decomposed) zones to convey state change.","palette":"Control-plane cool: deep slate rail #232B36, station field #E7EBEF, advance accent violet #6E56CF, gate-check green #2E9E6B. Levels L and XL rendered as violet route tags.","typography":"Strong geometric sans display for stage names with a narrow tabular companion for state indices and check counts. The machine/frame language drives heading weight.","desktopComposition":"The hero rail runs full-width horizontally near the top with four station markers; below, each section anchors to one rail station so scrolling walks the viewer along the state machine left-to-right.","mobileTransformation":"The horizontal rail divides into vertical station cards (Top→run→check) that stack as a reading list; the baseline turns into a left margin rule and each station becomes a tall step card instead of a wide marker row.","mediaStrategy":"No media region — assetManifest is empty. The state rail and module markers are a derived diagram of the verified state-machine claim, not a screenshot.","forbiddenPatterns":["dashboard/editor shell","three identical metric cards at hero","decorative glow used as visual idea","fake terminal or log output"],"compositionFingerprint":"stepped state rail with bolted station cards and an L/XL level tag traveling the track"},{"id":"level-contract-ledger","name":"Level Contract Ledger","projectMetaphor":"A decision ledger that sizes a task to L or XL, then tallies routed units and audits each delivered handoff — like the canonical module-execution.json contract being completed line by line.","signatureVisual":"A ruled ledger column where each routed working group is a line, running subtotals for units, and a final signed check row with pass marks — spelling out the 100→7→5→10→17 story as bookkeeping.","heroGeometry":"A multi-column ruled table geometry: three vertical ledger columns whose tops bind into a single header sum, then tally down to a bold signed total line with check glyphs.","sectionSequence":["ledger-hero-total","level-sizing-column","unit-tally-ledger","group-entries-rows","check-audit-column","vibe-reentry-contract"],"surfaceModel":"Paper ledger surfaces: hairline-ruled rows and columns on a soft stock, with a sewn binding edge, a heavier total rule, and a final wet-ink-like signature mark. No rounded-pill cards anywhere.","palette":"Financial-ink on aged stock: deep ink #1C1A17, parchment #F1ECE1, audit green #3E7C3E for passes, size-accent ocher #C99A3A for the L/XL line. Red rule slightly warm, not pure black.","typography":"Strong serif accounting voice for the ledger numbers (tabular lining figures) paired with a compact sans for group names — a bookkeeping hierarchy, not UI labels.","desktopComposition":"The hero is a true ruled ledger held at an angle with three columns visibly tallying downward; the case section re-uses the same ruled grammar so the 100→7→5→10→17 claim is read as one continuous audit trail.","mobileTransformation":"The three ledger columns fold into a single single-margin tally list with running subtotals; the multi-column audit rail becomes sequential ledger rows, keeping the ruling and the signed total rather than becoming cards.","mediaStrategy":"No media region — assetManifest is empty. The ruled ledger with its tally and signed total is a derived illustration of the verified audit-style counts, not a screenshot.","forbiddenPatterns":["terminal/console shell","three glowing stat bubbles","particle or grid background","fake JSON pretty-printed UI as screenshot"],"compositionFingerprint":"ruled accounting ledger with three tallying columns binding into a signed pass/total line"}],"selectedDirectionId":"routing-switchboard","selectionRationale":"Of the three, routing-switchboard most directly expresses the project's core mechanism (routing 100+ local Skills down to a chosen few routed groups per task) rather than its surrounding machinery. Its many-to-few convergence fan maps the single most concrete verified evidence — the ML case selecting 7 Skills from 100+ into 5 groups and 10 units — into the hero geometry, which planning-state-machine and level-contract-ledger only encode indirectly as pipeline or bookkeeping. It also best honors the 'no marketing hero and no feature-card wall' constraint: the visual idea is a routing topology derived from verified routing/grouping facts, not a generic hero. Its mobile transformation (fan→vertical taper spine) is a true recomposition rather than a column stack. All three avoid banned shells and none require media, since assetManifest is empty."},"designBrief":{"schemaVersion":"dodot-design-brief-v1","selectedDirectionId":"routing-switchboard","projectMetaphor":"A junction that takes many unlabeled incoming signals and dispatches each to exactly one wired lane — the essence of VibeSkills routing 100+ local Skills down to a chosen few routed modules per task.","signatureVisual":"A wire-fan junction: a dense bundle of thin signal lines (one per available local Skill) converges into a central switch hub, then fans out into fewer, thicker routed trunk lines labeled with the organized working groups. The density contrast (many-in, few-out) carries the routing claim.","compositionFingerprint":"many-to-few convergence fan with signal-thread rule-lines threading between translucent route-layer panels","heroGeometry":"A radial convergence fan: a wide arc of many fine lines drawn into a central trapezoid hub, then re-expanding into several bold routed paths with numbered pins.","sectionSequence":["convergence-hero","routing-mechanism-stage","signal-to-lane-mapping","case-breakdown-splits","delivery-track-results","install-condition"],"surfaceModel":"Signal planes: layered translucent paper-like panels that stack like routing layers, with thin connective rule-lines passing between panels as if wires thread seams. Sharp knuckle joints where lines change direction.","palette":"Ink on warm paper: near-black ink #16130F, warm cream field #F4EFE6, junction signal amber #E8A33D, routed trunk teal #1F6E63. Signals as amber threads, routed lanes as teal threads.","typography":"Monospaced display voice (thread/nodes feel) for headings paired with a humanist sans for routed body text. Numeric routing IDs rendered in tabular mono.","desktopComposition":"Left-anchored headline over the convergence fan on the right; the fan visually shrinks a field of fine lines into three routed trunk columns that carry the route→group→outcome story downward into the case section.","mobileTransformation":"The fan compresses into a vertical spine: fine input lines run down the left as a taper column, bottleneck into a single hub dot, then tilt into stacked routed lanes that read as scroll sequences rather than a side-by-side graph.","mediaStrategy":"No media region — assetManifest is empty and no local asset is safe. The signature wire-fan junction is a derived illustration of verified routing/group relationships, not a product screenshot.","evidencePlan":["hero convergence fan encodes the general-purpose routing claim (description) as many→few signal lines","routing-mechanism-stage renders the verified planning state machine / modular decomposition (readme claim) as the threading between route-layer panels","case-breakdown-splits + delivery-track-results render the ML case counts (100+ Skills → 7 selected → 5 groups → 10 units → 17/17 checks) as the fan's few bold trunk lines with number pins, never inventing extra metrics","install-condition renders the verified caveats (target skills directory, v4.0.0 vibe command re-entry) as the condition panel so no unsupported fact appears"],"localeContract":{"zh-CN":"中文页与英文页共用同一 brief、同一章节顺序、同一隐喻与构图；文案密度与措辞按语言自然化，路由数字（100+、7、5、10、17/17）与描述措辞均严格来自 material.json，不补充任何外部事实。","en":"The Chinese and English pages share the same brief, section order, metaphor and composition; copy density and wording are naturalized per language, and routing figures (100+, 7, 5, 10, 17/17) plus descriptive wording come strictly from material.json with no external facts added."},"forbiddenPatterns":["browser/terminal/editor shell","three-metric-card hero","radial gradient glow or particle field","fake product screenshot or placeholder frame","desktop fan merely stacked on mobile"]},"visualReview":{"schemaVersion":"dodot-automated-design-review-v1","status":"approved","summary":"双语文档忠实实现 routing-switchboard 设计方向。Hero 为真实存在的 many-to-few 收敛线扇（细琥珀信号线汇聚入中央梯形枢纽，再以粗 teal 干线扇出并标注 5 组/10 单元/17/17），六节顺序与 brief 完全一致，全部路由数字与描述均来自 material.json 的可验证事实，未补充任何外部指标或截图。手机端媒体查询将网格/线束重构为竖直阅读，是真实重组而非单纯堆列。英文与中文页结构、隐喻、配色、SVG 完全等价且措辞自然。无终端/浏览器外壳、无通用三卡 Hero、无装饰渐变网格、无假设备壳或假控件。仅存在一处次要清理建议（未使用的 .checks/.checkchip/.cf-note CSS），不构成 blocking。","checks":{"projectFit":{"passed":true,"evidence":"选中方向 routing-switchboard 被完整落实：hero 收敛线扇、palette(ink #16130F / cream #F4EFE6 / amber #E8A33D / teal #1F6E63)、mono 标题 + 人文等宽正文均在 zh-CN.html 与 en.html 中逐一对齐；sectionSequence 六节(convergence-hero,routing-mechanism-stage,signal-to-lane-mapping,case-breakdown-splits,delivery-track-results,install-condition)全部出现且顺序一致。"},"evidenceBound":{"passed":true,"evidence":"hero 扇面/pressrule/case 节的 100+、7、5、10、17/17 均能对应 summary.evidence 中的 ML 案例声明；L/XL 规模判定、module-execution.json、v4.0.0 vibe 重入、移除旧入口、~/.agents/skills 目录、不安装 chrome/playwright/context7/claude-flow 均为 summary 中可验证的 readme/release 声明；未见额外杜撰指标。"},"signatureVisual":{"passed":true,"evidence":"两页内联 SVG(.fan)真实存在：多条细 amber 输入线(100+ Skills)汇聚进黑色梯形枢纽(KEY/根)，再以 3 条粗 teal 干线扇出，并标注 '5 groups/5 组'、'10 units/10 单元'、'17/17'，正是 brief 要求的 many-to-few 收敛线扇签名视觉。"},"mobileTransformation":{"passed":true,"evidence":"@media(max-width:820px) 将 hero-grid/grid2/grid3 折叠为单列、pressrule 改纵向、fan-wrap 上移(order:-1)成为纵向管道、thrumap 变竖排且节点从 border-left 改为 border-top 以重定向阅读方向，符合 brief 中“收敛扇压缩为纵向脊柱”的重构而非单纯堆叠列，QA 移动端 scrollWidth=390 无横向溢出。"},"localeParity":{"passed":true,"evidence":"zh-CN.html(lang=zh-CN)与 en.html(lang=en)共用同一结构、六节顺序、相同 SVG、相同配色与数字；中文文案密度自然化(如中文 1447 字符/英文 2910 字符反映语言特性)，数值与描述措辞均来自 material.json，无外部事实；QA 中两语言 h1Count=1/h2Count=5 一致。"},"antiTemplate":{"passed":true,"evidence":"未回退到通用 Hero/三卡墙/卡片墙；hero 主体是项目专属的路由收敛扇而非通用营销视觉，案例节复用同套扇状条形语法讲故事；无非依据的设备外壳、无装饰性径向渐变/粒子网格(背景为纯 crea cream 平色)。"},"noFakeUi":{"passed":true,"evidence":"无终端/浏览器/编辑器外壳，无伪造产品截图或占位框，无伪命令日志输出；module-execution.json 仅作等宽 code 样式引用，无真实控件；QA 报告 fakeControlCount=0 佐证。"}},"issues":[{"code":"minor-unused-css","severity":"warning","locale":"all","description":".checks/.checkchip/.cf-note 等 CSS 规则在当前 DOM 中未被使用，属死代码，不影响渲染与断言。","repairInstruction":"可选：从两页样式块中移除未使用的 .checks/.checkchip/.cf-note，使 CSS 与实际渲染一一对应；不作为发布阻断项。"}]},"repairEvidence":[],"provenance":{"executor":"opencode","model":"deepseek/deepseek-v4-flash","opencodeVersion":"1.1.58","promptVersion":"dodot-showcase-deepseek-v4","inputHash":"dce5f565ae8687d808484f76eaaa227efb2db3bb483f5356e6e3e4736323d1f2","artifactHash":"b85d27c1200d0627289cdd58f67dd47e0c681e12c5d9cacac732cf822098cefa","designSkill":"dodot-showcase-design"},"qaEvidence":{"zh-CN":[{"viewport":"desktop","passed":true,"localePassed":true,"metrics":{"language":"zh-CN","h1Count":1,"h2Count":5,"sectionCount":5,"textLength":1447,"cjkCount":622,"latinLetterCount":401,"scrollWidth":1280,"clientWidth":1280,"scrollHeight":3834,"h1Top":138.796875,"bottomInset":277.3125,"brokenImageCount":0,"missingAltCount":0,"fakeControlCount":0,"emptyLinkCount":0},"screenshotHashes":{"top":"1e8c950c00132aac821cdff70e3a4ddba170767ed4d134de0049aca7174c3080","middle":"474967b4157c9bf02e71e12fa71933c9fe39481fb2e0a7fb710fde15ad959c73","bottom":"922b1c954f4c1f71a3635ad0d4a08de1f5fcd7122aa68e96bf332055858dd357"},"consoleErrors":[],"networkRequests":[]},{"viewport":"mobile","passed":true,"localePassed":true,"metrics":{"language":"zh-CN","h1Count":1,"h2Count":5,"sectionCount":5,"textLength":1447,"cjkCount":622,"latinLetterCount":401,"scrollWidth":390,"clientWidth":390,"scrollHeight":5178,"h1Top":344.921875,"bottomInset":354.015625,"brokenImageCount":0,"missingAltCount":0,"fakeControlCount":0,"emptyLinkCount":0},"screenshotHashes":{"top":"40bcd982ed50a0cc6d2c57b307b96823ec18e1ad1d9261deb25c5211bc65d498","middle":"a7cc056405bacc09db860395dfce7425eb087cb5ccf488b62c795bef6e624932","bottom":"75c8f5df7f523737f7be0819859a2676fb3b63a0524f7f129354d37baf5ab8c4"},"consoleErrors":[],"networkRequests":[]}],"en":[{"viewport":"desktop","passed":true,"localePassed":true,"metrics":{"language":"en","h1Count":1,"h2Count":5,"sectionCount":5,"textLength":2910,"cjkCount":0,"latinLetterCount":2193,"scrollWidth":1280,"clientWidth":1280,"scrollHeight":4118,"h1Top":138.796875,"bottomInset":317.71875,"brokenImageCount":0,"missingAltCount":0,"fakeControlCount":0,"emptyLinkCount":0},"screenshotHashes":{"top":"b16ff7cca7b2e9ed51e1190bcab313d38a4662335f30f0ee3fcc87478934e0b5","middle":"83bd2ca706a39164fc61c1fdf0cd8ef3c06fe943ca348fbb5ffd9e902c02e9eb","bottom":"d26857efea8800f813f5bdec75b790b518b53086c5f7a30845352911179981af"},"consoleErrors":[],"networkRequests":[]},{"viewport":"mobile","passed":true,"localePassed":true,"metrics":{"language":"en","h1Count":1,"h2Count":5,"sectionCount":5,"textLength":2910,"cjkCount":0,"latinLetterCount":2193,"scrollWidth":390,"clientWidth":390,"scrollHeight":5714,"h1Top":365.71875,"bottomInset":354.671875,"brokenImageCount":0,"missingAltCount":0,"fakeControlCount":0,"emptyLinkCount":0},"screenshotHashes":{"top":"abd5e697616373d3a5bf0d6d83ffe0e45efd571d1ae728093b98623e1c944edb","middle":"c610c0e77568b79bbfbc63829f6bb5e38969fd599f2bd82958877b02d2902950","bottom":"e86d0e4fb06c3e372233503f3360204de557c5bdb40538c919d142160697104f"},"consoleErrors":[],"networkRequests":[]}]}},"generatorVersion":"opencode-1.1.58:deepseek/deepseek-v4-flash:dodot-showcase-deepseek-v4","publishedAt":"2026-08-17T15:24:38.473Z"},"rawExtraction":{"projectRevisionId":"cmt2zh06r00mn62716aoj8asc","projectVersion":1,"sources":[{"canonicalKey":"github:foryourhealth111-pixel/vibe-skills","role":"primary","revisionVersion":1,"url":"https://github.com/foryourhealth111-pixel/Vibe-Skills"}],"facts":[{"key":"repository_owner","value":"foryourhealth111-pixel","evidence":[{"path":"repository.owner","excerpt":"foryourhealth111-pixel"}]},{"key":"repository_name","value":"vibe-skills","evidence":[{"path":"repository.name","excerpt":"vibe-skills"}]},{"key":"description","value":"VibeSkills is a general-purpose Skill that automatically routes local Skills and intelligently orchestrates harness workflows.","evidence":[{"path":"repository.description","excerpt":"VibeSkills is a general-purpose Skill that automatically routes local Skills and intelligently orchestrates harness workflows."}]},{"key":"primary_language","value":"Python","evidence":[{"path":"repository.language","excerpt":"Python"}]},{"key":"license","value":"Apache-2.0","evidence":[{"path":"repository.license","excerpt":"Apache-2.0"}]},{"key":"topics","value":"agent-framework, agent-skills, agentic-coding, ai-agents, ai-scientist, ai-skills, ai-workflow, automation, claude-code, codex, context-engineering, developer-tools, dsh-plugin, llm, multi-agent, prompt-engineering, skills, vibe-coding, vibecoding, workflow-automation","evidence":[{"path":"repository.topics","excerpt":"agent-framework, agent-skills, agentic-coding, ai-agents, ai-scientist, ai-skills, ai-workflow, automation, claude-code, codex, context-engineering, developer-tools, dsh-plugin, llm, multi-agent, prompt-engineering, skills, vibe-coding, vibecoding, workflow-automation"}]},{"key":"source_count","value":"1","evidence":[{"path":"projectRevision.sources","excerpt":"github:foryourhealth111-pixel/vibe-skills"}]}],"taxonomy":{"category":"development_tools","subcategory":"data_ai_tools","confidence":0.65,"evidence":[{"path":"repository.description","excerpt":"VibeSkills is a general-purpose Skill that automatically routes local Skills and intelligently orchestrates harness workflows."}]},"lifecycle":{"stage":"active","confidence":0.75,"evidence":[{"path":"repository.pushedAt","excerpt":"2026-08-11T15:42:32.000Z"}]},"userIntents":[{"slug":"build_software","label":"构建或改进软件","confidence":0.65,"evidence":[{"path":"repository.description","excerpt":"VibeSkills is a general-purpose Skill that automatically routes local Skills and intelligently orchestrates harness workflows."}]},{"slug":"automate_work","label":"自动化工作流程","confidence":0.65,"evidence":[{"path":"repository.description","excerpt":"VibeSkills is a general-purpose Skill that automatically routes local Skills and intelligently orchestrates harness workflows."}]},{"slug":"create_content","label":"创作内容或设计","confidence":0.65,"evidence":[{"path":"repository.description","excerpt":"VibeSkills is a general-purpose Skill that automatically routes local Skills and intelligently orchestrates harness workflows."}]},{"slug":"analyze_data","label":"分析与理解数据","confidence":0.65,"evidence":[{"path":"repository.description","excerpt":"VibeSkills is a general-purpose Skill that automatically routes local Skills and intelligently orchestrates harness workflows."}]},{"slug":"learn_explore","label":"学习与探索","confidence":0.65,"evidence":[{"path":"repository.description","excerpt":"VibeSkills is a general-purpose Skill that automatically routes local Skills and intelligently orchestrates harness workflows."}]}],"readme":{"headings":[],"commands":[]},"changeSet":[{"field":"repository","beforeHash":"1a7a3f9ef20e78cfb0182ca274db810af1a99d8121889bcdca733ab04c86a5bf","afterHash":"7864f2d006fec7658010e88c2baf0738ab4effff1b3dcd22b23e297bac7730e5"}]},"publishedAt":"2026-08-10T15:20:45.280Z","updatedAt":"2026-08-21T13:26:50.183Z","summary":{"id":"cmt2zh06u00mp6271pakzl1xr","projectType":"agent","oneLine":"通用型 Skill，自动路由并编排本地 Skills 与 harness 工作流。","overview":"VibeSkills 是一个通用型 Skill，自动路由本地 Skills 并智能编排 harness 工作流。它遵循类似 Superpower 和 GSD 的框架，通过规划状态机进行模块化分解，用不同 Skills 辅助不同模块，减少用户的规划和认知负担。在机器学习实验案例中，它从 100+ Skills 中选出 7 个，组织成 5 个工作组和 10 个工作单元，最终 17/17 项检查全部通过，提供端到端的交付体验。","highlights":["自动路由本地 Skills","智能编排 harness 工作流","模块化分解减少认知负担","端到端交付体验"],"bestFor":["希望自动管理多个本地 Skills 的开发者","需要复杂任务编排的 AI Agent 用户","追求端到端交付体验的 Vibe Coding 实践者"],"caveats":["需要安装到特定 Skill 目录（如 ~/.agents/skills）","v4.0.0 移除了旧入口，需使用 vibe 命令","案例展示的是特定场景，实际效果可能因环境而异"],"categorySuggestion":"development_tools","subcategorySuggestion":"coding_agents","lifecycleStages":["idea_requirements","product_design","testing_quality"],"audienceIntents":["discover","learn","build","integrate"],"confidence":0.85,"evidence":[{"claim":"VibeSkills 是一个通用型 Skill，自动路由本地 Skills 并智能编排 harness 工作流。","sourceType":"description","locator":"https://github.com/foryourhealth111-pixel/Vibe-Skills"},{"claim":"项目核心是遵循类似 Superpower 和 GSD 的 harness 框架，基于规划状态机进行模块化分解，用不同 Skills 辅助不同模块，减少用户的规划和认知负担。","sourceType":"readme","locator":"https://github.com/foryourhealth111-pixel/Vibe-Skills"},{"claim":"在机器学习实验案例中，VibeSkills 从 100+ Skills 中选出 7 个，组织成 5 个工作组和 10 个工作单元，最终 17/17 项检查全部通过。","sourceType":"readme","locator":"https://github.com/foryourhealth111-pixel/Vibe-Skills"},{"claim":"VibeSkills 为 Agent 提供从接收任务到检查交付的完整流程，包括确认需求、推荐 L 或 XL 级别、组织 Skills、记录工作和检查结果。","sourceType":"readme","locator":"https://github.com/foryourhealth111-pixel/Vibe-Skills"},{"claim":"v4.0.0 是主要的 Agent 执行交接版本，要求通过 vibe 重新入口返回规范的 module-execution.json，并移除了旧的公开入口名称。","sourceType":"release","locator":"https://github.com/foryourhealth111-pixel/Vibe-Skills/releases/tag/v4.0.0"},{"claim":"v4.0.0 不安装或推荐 chrome、chrome-devtools、playwright、context7 或 claude-flow MCP。","sourceType":"release","locator":"https://github.com/foryourhealth111-pixel/Vibe-Skills/releases/tag/v4.0.0"}],"publishedAt":"2026-08-17T13:23:12.885Z"},"github":{"url":"https://github.com/foryourhealth111-pixel/Vibe-Skills","ownerAvatarUrl":"https://avatars.githubusercontent.com/u/251585801?v=4","starsCount":2869,"forksCount":228,"language":"Python","syncedAt":"2026-08-17T13:18:58.393Z","license":"Apache-2.0","defaultBranch":"main","archived":false,"disabled":false,"lastChangedAt":"2026-08-11T15:42:32.000Z","missing":false},"tags":[{"name":"Vibe coding","slug":"vibe-coding","verification":"verified","evidenceUrl":"https://github.com/foryourhealth111-pixel/Vibe-Skills","evidenceExcerpt":"GitHub topic: vibe-coding"}],"updates":[]},"entryPoint":"discovery","claim":{"state":"unclaimed","verified":false},"initialView":"beautified","community":{"viewer":null,"state":{"likesCount":0,"creativeCount":0,"followersCount":0,"commentsCount":0,"liked":false,"creative":false,"followed":false},"comments":{"comments":[],"nextCursor":null}}}]}],"$L24"]