[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M736SPHSAYMW5XVHRBF76163":3,"skill-same:01M736SPHSAYMW5XVHRBF76163":58},{"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":28,"updatedAt":29,"install":30,"resources":43},"01M736SPHSAYMW5XVHRBF76163","guided-feature-build",null,"Run a guided, in-session feature build on an existing codebase — explore, design, implement, review — without opening issue-backed plan tracking.","graysurf\u002Fagent-runtime-kit","graysurf","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F10785178?v=4","01M736PQHYKHFMNBYF8BT93NM4","01M736PQJWYH7G046WAN7EQCHD",0,"MIT","https:\u002F\u002Fgithub.com\u002Fgraysurf\u002Fagent-runtime-kit\u002Fblob\u002Fmain\u002Ftests\u002Fgolden\u002Fcodex\u002Fplugins\u002Fconversation\u002Fskills\u002Fguided-feature-build\u002Fexpected\u002FSKILL.md",[],[],"2026-11-11 15:11:14",false,"---\nname: guided-feature-build\ndescription: >\n  Run a guided, in-session feature build on an existing codebase — explore,\n  design, implement, review — without opening issue-backed plan tracking.\n---\n\n# Guided Feature Build\n\n## Contract\n\nPrereqs:\n\n- User explicitly invokes `guided-feature-build`, asks for a guided feature\n  build, or asks to be walked through building a feature before any code is\n  written.\n- The delegated role prompts exist at `references\u002Fprompts\u002Fexplorer.md` and\n  `references\u002Fprompts\u002Farchitect.md`.\n- The internal execution strategy is available at\n  `references\u002FDELEGATION_PROTOCOL.md`.\n- Active project preflight and validation rules are followed before any\n  repository edits.\n\nInputs:\n\n- Feature request, the problem it solves, constraints, target area, done\n  criteria, and any preferred approach when available.\n- Whether subagent delegation is available and allowed by active runtime\n  instructions before any subagents are spawned.\n- Any explicit delegation preference or constraint already stated by the user.\n\nOutputs:\n\n- A `TodoWrite` plan covering the seven phases.\n- A codebase-exploration summary with the key files that were read.\n- A clarifying-question list answered by the user before design.\n- Two or three compared architecture approaches with one recommendation and the\n  user's chosen approach.\n- An implementation that follows the chosen approach and codebase conventions,\n  validated per project rules.\n- A quality-review pass (delegated to the existing code-review skills) and a\n  closing summary of what was built, decisions, files changed, and next steps.\n\nExit codes:\n\n- N\u002FA (conversation workflow; no repo scripts)\n\nFailure modes:\n\n- A delegated role prompt is missing or empty.\n- The request is a trivial change, a hotfix, or already fully specified — use\n  direct implementation instead of this conductor.\n- The work needs durable issue tracking, multiple delivery lanes, or PR\n  grouping — hand off to the plan-tracking or dispatch workflow instead.\n- Active project rules conflict with the requested implementation path.\n\n## Outcome Routing\n\nAdvice and explanation stay normal conversation behavior: ask only material\nclarifying questions, state assumptions, compare meaningful options, and give\none recommendation without asking the user to select a prompt-style skill.\n\nFor an implementation outcome, this workflow selects inline, orchestrated, or\nparallel execution internally using `references\u002FDELEGATION_PROTOCOL.md`.\nDelegation is an implementation detail, not a separate user outcome. The\nworkflow escalates to plan tracking or dispatch only when durable coordination\nstate is actually required.\n\n## Workflow\n\nThis skill conducts the build through fixed phases with explicit user gates. It\ndelegates exploration and architecture to subagents when delegation is\navailable, and degrades to inline sequential passes when it is not. It does not\nre-implement review: phase 6 calls the existing code-review skills.\n\n### Phase 1 — Discovery\n\n1. Restate the feature request and create a `TodoWrite` list with all seven\n   phases.\n2. If the request is unclear, ask what problem it solves, what the feature\n   should do, and any constraints. Confirm the understanding before proceeding.\n\n### Phase 2 — Codebase Exploration\n\n1. If subagent delegation is available, dispatch 2–3 subagents using\n   `references\u002Fprompts\u002Fexplorer.md`, each targeting a different aspect (similar\n   features, architecture\u002Fabstractions, the current implementation of a related\n   area). If delegation is unavailable, run the same role prompt inline as\n   sequential read passes.\n2. Each pass returns entry points with `file:line` references and a curated list\n   of 5–10 key files to read.\n3. Read every file the passes flag before continuing, then present a concise\n   summary of the patterns and constraints discovered.\n\n### Phase 3 — Clarifying Questions\n\nDo not skip this gate.\n\n1. From the exploration findings and the original request, list every\n   underspecified aspect: edge cases, error handling, integration points, scope\n   boundaries, backward compatibility, and performance needs.\n2. Present the questions as one organized list and **wait for answers** before\n   designing. If the user defers (\"whatever you think is best\"), state your\n   recommended answers and get explicit confirmation.\n\n### Phase 4 — Architecture Design\n\n1. If delegation is available, dispatch 2–3 subagents using\n   `references\u002Fprompts\u002Farchitect.md` with distinct focuses — minimal change,\n   clean architecture, and pragmatic balance. Degrade to inline sequential\n   passes when delegation is unavailable.\n2. Compare the approaches, form one recommendation with reasoning, and present\n   the trade-offs.\n3. **Ask the user which approach to use** and wait for the choice.\n\n### Phase 5 — Implementation\n\nDo not start without explicit user approval.\n\n1. Wait for approval of the chosen approach.\n2. Select the execution strategy through\n   `references\u002FDELEGATION_PROTOCOL.md`. Honor an explicit user preference when\n   safe, but keep scope, write isolation, integration, and validation under the\n   parent workflow.\n3. For a testable production behavior change, follow the policy-owned\n   durable test-first lifecycle: contract delta, affected-test scan, meaningful\n   red or complete waiver before production edits, scoped validation, suite\n   convergence, and explicit residual gaps. Record v2 evidence through the\n   `test-first-evidence` CLI; it is required when the repo or user opts into the\n   `forge-cli` `[test_first].require` gate.\n4. Read the relevant files again, implement the chosen approach, follow codebase\n   conventions strictly, and keep `TodoWrite` current.\n5. Run the project's required preflight and validation as edits land.\n\n### Phase 6 — Quality Review\n\n1. Route the diff to the generic `code-review-specialists` outcome. It selects\n   ad-hoc, follow-up, or pre-merge context and chooses quick, focused, or\n   specialist depth from the request and risk. Do not define a separate review\n   rubric here.\n2. Present the consolidated findings and ask the user how to proceed — fix now,\n   fix later, or proceed as-is — then act on the decision. When this build is\n   being delivered through a PR\u002FMR, surface the findings on it before fixing,\n   never after, per the posting-order invariant in\n   `skills\u002Fcode-review\u002Fcode-review-specialists\u002Freferences\u002FREVIEW_OUTCOME_POSTING_CONTRACT.md`.\n\n### Phase 7 — Summary\n\n1. Mark the todos complete.\n2. Summarize what was built, the key decisions, the files changed, and suggested\n   next steps.\n3. If the work now needs durable tracking, multiple lanes, or PR grouping, it\n   has outgrown this conductor — stop here and let the user choose the tracking\n   workflow rather than continuing.\n\n## Boundary\n\n- This is an in-session conductor. It creates no provider issues, plans, or\n  tracking artifacts; on escalation it stops and hands the work to the user's\n  chosen tracking workflow (Phase 7).\n- It ships as a cross-product skill only. It defines no `agents\u002F\u003Cn>.md` and no\n  `commands\u002F\u003Cn>.md`; the explorer and architect roles live as delegated role\n  prompts under `references\u002Fprompts\u002F`, fulfilled by each harness's own\n  delegation capability.\n- It reuses the generic code-review outcome for Phase 6, the internal delegation\n  protocol, and the policy-owned durable test-first lifecycle for Phase 5\n  instead of exposing those mechanics as user-selected substeps.\n","# Guided Feature Build\n\n## Contract\n\nPrereqs:\n\n- User explicitly invokes `guided-feature-build`, asks for a guided feature\n  build, or asks to be walked through building a feature before any code is\n  written.\n- The delegated role prompts exist at `references\u002Fprompts\u002Fexplorer.md` and\n  `references\u002Fprompts\u002Farchitect.md`.\n- The internal execution strategy is available at\n  `references\u002FDELEGATION_PROTOCOL.md`.\n- Active project preflight and validation rules are followed before any\n  repository edits.\n\nInputs:\n\n- Feature request, the problem it solves, constraints, target area, done\n  criteria, and any preferred approach when available.\n- Whether subagent delegation is available and allowed by active runtime\n  instructions before any subagents are spawned.\n- Any explicit delegation preference or constraint already stated by the user.\n\nOutputs:\n\n- A `TodoWrite` plan covering the seven phases.\n- A codebase-exploration summary with the key files that were read.\n- A clarifying-question list answered by the user before design.\n- Two or three compared architecture approaches with one recommendation and the\n  user's chosen approach.\n- An implementation that follows the chosen approach and codebase conventions,\n  validated per project rules.\n- A quality-review pass (delegated to the existing code-review skills) and a\n  closing summary of what was built, decisions, files changed, and next steps.\n\nExit codes:\n\n- N\u002FA (conversation workflow; no repo scripts)\n\nFailure modes:\n\n- A delegated role prompt is missing or empty.\n- The request is a trivial change, a hotfix, or already fully specified — use\n  direct implementation instead of this conductor.\n- The work needs durable issue tracking, multiple delivery lanes, or PR\n  grouping — hand off to the plan-tracking or dispatch workflow instead.\n- Active project rules conflict with the requested implementation path.\n\n## Outcome Routing\n\nAdvice and explanation stay normal conversation behavior: ask only material\nclarifying questions, state assumptions, compare meaningful options, and give\none recommendation without asking the user to select a prompt-style skill.\n\nFor an implementation outcome, this workflow selects inline, orchestrated, or\nparallel execution internally using `references\u002FDELEGATION_PROTOCOL.md`.\nDelegation is an implementation detail, not a separate user outcome. The\nworkflow escalates to plan tracking or dispatch only when durable coordination\nstate is actually required.\n\n## Workflow\n\nThis skill conducts the build through fixed phases with explicit user gates. It\ndelegates exploration and architecture to subagents when delegation is\navailable, and degrades to inline sequential passes when it is not. It does not\nre-implement review: phase 6 calls the existing code-review skills.\n\n### Phase 1 — Discovery\n\n1. Restate the feature request and create a `TodoWrite` list with all seven\n   phases.\n2. If the request is unclear, ask what problem it solves, what the feature\n   should do, and any constraints. Confirm the understanding before proceeding.\n\n### Phase 2 — Codebase Exploration\n\n1. If subagent delegation is available, dispatch 2–3 subagents using\n   `references\u002Fprompts\u002Fexplorer.md`, each targeting a different aspect (similar\n   features, architecture\u002Fabstractions, the current implementation of a related\n   area). If delegation is unavailable, run the same role prompt inline as\n   sequential read passes.\n2. Each pass returns entry points with `file:line` references and a curated list\n   of 5–10 key files to read.\n3. Read every file the passes flag before continuing, then present a concise\n   summary of the patterns and constraints discovered.\n\n### Phase 3 — Clarifying Questions\n\nDo not skip this gate.\n\n1. From the exploration findings and the original request, list every\n   underspecified aspect: edge cases, error handling, integration points, scope\n   boundaries, backward compatibility, and performance needs.\n2. Present the questions as one organized list and **wait for answers** before\n   designing. If the user defers (\"whatever you think is best\"), state your\n   recommended answers and get explicit confirmation.\n\n### Phase 4 — Architecture Design\n\n1. If delegation is available, dispatch 2–3 subagents using\n   `references\u002Fprompts\u002Farchitect.md` with distinct focuses — minimal change,\n   clean architecture, and pragmatic balance. Degrade to inline sequential\n   passes when delegation is unavailable.\n2. Compare the approaches, form one recommendation with reasoning, and present\n   the trade-offs.\n3. **Ask the user which approach to use** and wait for the choice.\n\n### Phase 5 — Implementation\n\nDo not start without explicit user approval.\n\n1. Wait for approval of the chosen approach.\n2. Select the execution strategy through\n   `references\u002FDELEGATION_PROTOCOL.md`. Honor an explicit user preference when\n   safe, but keep scope, write isolation, integration, and validation under the\n   parent workflow.\n3. For a testable production behavior change, follow the policy-owned\n   durable test-first lifecycle: contract delta, affected-test scan, meaningful\n   red or complete waiver before production edits, scoped validation, suite\n   convergence, and explicit residual gaps. Record v2 evidence through the\n   `test-first-evidence` CLI; it is required when the repo or user opts into the\n   `forge-cli` `[test_first].require` gate.\n4. Read the relevant files again, implement the chosen approach, follow codebase\n   conventions strictly, and keep `TodoWrite` current.\n5. Run the project's required preflight and validation as edits land.\n\n### Phase 6 — Quality Review\n\n1. Route the diff to the generic `code-review-specialists` outcome. It selects\n   ad-hoc, follow-up, or pre-merge context and chooses quick, focused, or\n   specialist depth from the request and risk. Do not define a separate review\n   rubric here.\n2. Present the consolidated findings and ask the user how to proceed — fix now,\n   fix later, or proceed as-is — then act on the decision. When this build is\n   being delivered through a PR\u002FMR, surface the findings on it before fixing,\n   never after, per the posting-order invariant in\n   `skills\u002Fcode-review\u002Fcode-review-specialists\u002Freferences\u002FREVIEW_OUTCOME_POSTING_CONTRACT.md`.\n\n### Phase 7 — Summary\n\n1. Mark the todos complete.\n2. Summarize what was built, the key decisions, the files changed, and suggested\n   next steps.\n3. If the work now needs durable tracking, multiple lanes, or PR grouping, it\n   has outgrown this conductor — stop here and let the user choose the tracking\n   workflow rather than continuing.\n\n## Boundary\n\n- This is an in-session conductor. It creates no provider issues, plans, or\n  tracking artifacts; on escalation it stops and hands the work to the user's\n  chosen tracking workflow (Phase 7).\n- It ships as a cross-product skill only. It defines no `agents\u002F\u003Cn>.md` and no\n  `commands\u002F\u003Cn>.md`; the explorer and architect roles live as delegated role\n  prompts under `references\u002Fprompts\u002F`, fulfilled by each harness's own\n  delegation capability.\n- It reuses the generic code-review outcome for Phase 6, the internal delegation\n  protocol, and the policy-owned durable test-first lifecycle for Phase 5\n  instead of exposing those mechanics as user-selected substeps.\n",{"name":5,"description":7},"main","da7119f2ebece510d6955cea72743069f5496c73","tests\u002Fgolden\u002Fcodex\u002Fplugins\u002Fconversation\u002Fskills\u002Fguided-feature-build\u002Fexpected","https:\u002F\u002Fgithub.com\u002Fgraysurf\u002Fagent-runtime-kit","https:\u002F\u002Fgithub.com\u002Fgraysurf","Building reliable tools for teams of AI coding agents · Rust \u002F TypeScript · @sympoies","2026-08-29 05:27:47",{"source":8,"skillName":5,"commands":31,"prompt":41,"zipAvailable":42},[32,35,38],{"tool":33,"command":34},"npx","npx skills add graysurf\u002Fagent-runtime-kit --skill guided-feature-build",{"tool":36,"command":37},"bunx","bunx skills add graysurf\u002Fagent-runtime-kit --skill guided-feature-build",{"tool":39,"command":40},"pnpm","pnpm dlx skills add graysurf\u002Fagent-runtime-kit --skill guided-feature-build","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fgraysurf\u002Fagent-runtime-kit\u002Fblob\u002Fmain\u002Ftests\u002Fgolden\u002Fcodex\u002Fplugins\u002Fconversation\u002Fskills\u002Fguided-feature-build\u002Fexpected\u002FSKILL.md",true,[44,50,54],{"path":45,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":48,"sourceHtmlUrl":49},"references\u002FDELEGATION_PROTOCOL.md","REFERENCE","参考文档",2103,"https:\u002F\u002Fgithub.com\u002Fgraysurf\u002Fagent-runtime-kit\u002Fblob\u002Fmain\u002Ftests\u002Fgolden\u002Fcodex\u002Fplugins\u002Fconversation\u002Fskills\u002Fguided-feature-build\u002Fexpected\u002Freferences\u002FDELEGATION_PROTOCOL.md",{"path":51,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":52,"sourceHtmlUrl":53},"references\u002Fprompts\u002Farchitect.md",1554,"https:\u002F\u002Fgithub.com\u002Fgraysurf\u002Fagent-runtime-kit\u002Fblob\u002Fmain\u002Ftests\u002Fgolden\u002Fcodex\u002Fplugins\u002Fconversation\u002Fskills\u002Fguided-feature-build\u002Fexpected\u002Freferences\u002Fprompts\u002Farchitect.md",{"path":55,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":56,"sourceHtmlUrl":57},"references\u002Fprompts\u002Fexplorer.md",1275,"https:\u002F\u002Fgithub.com\u002Fgraysurf\u002Fagent-runtime-kit\u002Fblob\u002Fmain\u002Ftests\u002Fgolden\u002Fcodex\u002Fplugins\u002Fconversation\u002Fskills\u002Fguided-feature-build\u002Fexpected\u002Freferences\u002Fprompts\u002Fexplorer.md",[59,65,70,77,84,91],{"skillNo":60,"name":61,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":62,"categories":63,"keywords":64,"publishedAt":18},"01M736XWK14DFRERMQ8RT3HS0Z","removable-skill","https:\u002F\u002Fgithub.com\u002Fgraysurf\u002Fagent-runtime-kit\u002Fblob\u002Fmain\u002Ftests\u002Fruntime-smoke\u002Ffixtures\u002Fskill-lifecycle\u002Fremove-skill\u002Ftests\u002Fgolden\u002Fcodex\u002Fplugins\u002Ffixture\u002Fskills\u002Fremovable-skill\u002FSKILL.md",[],[],{"skillNo":66,"name":61,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":67,"categories":68,"keywords":69,"publishedAt":18},"01M736XTPQEKAYESDM1REN2KAN","https:\u002F\u002Fgithub.com\u002Fgraysurf\u002Fagent-runtime-kit\u002Fblob\u002Fmain\u002Ftests\u002Fruntime-smoke\u002Ffixtures\u002Fskill-lifecycle\u002Fremove-skill\u002Ftests\u002Fgolden\u002Fclaude\u002Fplugins\u002Ffixture\u002Fskills\u002Fremovable-skill\u002FSKILL.md",[],[],{"skillNo":71,"name":72,"nameZh":6,"description":73,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":74,"categories":75,"keywords":76,"publishedAt":18},"01M736XRMWHMJB4X563G9J6QYY","removable-project-skill","Fixture project-local skill used to prove remove-project-skill dry-run coverage.","https:\u002F\u002Fgithub.com\u002Fgraysurf\u002Fagent-runtime-kit\u002Fblob\u002Fmain\u002Ftests\u002Fruntime-smoke\u002Ffixtures\u002Fskill-lifecycle\u002Fremove-project-skill\u002F.agents\u002Fskills\u002Fremovable-project-skill\u002FSKILL.md",[],[],{"skillNo":78,"name":79,"nameZh":6,"description":80,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":81,"categories":82,"keywords":83,"publishedAt":18},"01M736XPT65YZMPCBD4K3RHV4P","project-sample-skill","Fixture project-local skill used to prove create-project-skill governance coverage.","https:\u002F\u002Fgithub.com\u002Fgraysurf\u002Fagent-runtime-kit\u002Fblob\u002Fmain\u002Ftests\u002Fruntime-smoke\u002Ffixtures\u002Fskill-lifecycle\u002Fcreate-project-skill\u002F.agents\u002Fskills\u002Fproject-sample-skill\u002FSKILL.md",[],[],{"skillNo":85,"name":86,"nameZh":6,"description":87,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":88,"categories":89,"keywords":90,"publishedAt":18},"01M736XKZVYWVQA1YB0P459WND","project-local-skill","Fixture project-local skill used by runtime-kit project-local smoke checks.","https:\u002F\u002Fgithub.com\u002Fgraysurf\u002Fagent-runtime-kit\u002Fblob\u002Fmain\u002Ftests\u002Fprojects\u002Fproject-local-smoke\u002F.agents\u002Fskills\u002Fproject-local-skill\u002FSKILL.md",[],[],{"skillNo":92,"name":93,"nameZh":6,"description":94,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":95,"categories":96,"keywords":97,"publishedAt":18},"01M736XHA73KZ5G2TKWSED5P1B","topic-radar","Aggregate read-only AI and technology trend signals (Polymarket, Hacker News, GitHub, arXiv, Hugging Face, news search) into source-grounded Markdown or JSON; the lower-level data engine behind `daily-brief`.","https:\u002F\u002Fgithub.com\u002Fgraysurf\u002Fagent-runtime-kit\u002Fblob\u002Fmain\u002Ftests\u002Fgolden\u002Fhermes\u002Fplugins\u002Freporting\u002Fskills\u002Ftopic-radar\u002Fexpected\u002FSKILL.md",[],[]]