[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M75Y2Q86FDGWQDJYYVC3F11M":3,"skill-same:01M75Y2Q86FDGWQDJYYVC3F11M":45},{"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":24,"commitSha":25,"skillDirectory":26,"forksCount":27,"repositoryHtmlUrl":28,"repositoryHomepageUrl":29,"ownerHtmlUrl":30,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":31,"install":32,"resources":44},"01M75Y2Q86FDGWQDJYYVC3F11M","implement",null,"Start coding or resume the approved current feature or fix on its branch, testing each step and presenting review. Use for \u002Fimplement or building current-feature.md.","aiblueprinthq\u002Fai-blueprint","aiblueprinthq","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F317905365?v=4","01M75XZGKSB8GPVXQ69BA57VA7","01M75XZGME4HNZG9CDB8K3TQ8X",510,"MIT","https:\u002F\u002Fgithub.com\u002Faiblueprinthq\u002Fai-blueprint\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fimplement\u002FSKILL.md",[],[],"2026-11-12 13:55:43",false,"---\nname: implement\ndescription: Start coding or resume the approved current feature or fix on its branch, testing each step and presenting review. Use for \u002Fimplement or building current-feature.md.\ndisable-model-invocation: true\n---\n\n# implement - build the approved spec\n\n**Context reuse:** Reuse any required file already loaded in project instructions or the current session. Read it again only if absent, changed, or exact current bytes or line references are needed.\n\nThe approved `blueprint\u002Fcontext\u002Fcurrent-feature.md` is the authoritative feature\npacket. Implement it and stop before the work-level commit or merge.\n\n## Start\n\n**First action:** Before project inspection, preflight, or any other tool call,\npublish the `implement` activity as `running` when `blueprint\u002F.state\u002F` exists.\nFrom the project root, run the first helper that exists,\n`node .agents\u002Fskills\u002Fblueprint-doctor\u002Fscripts\u002Frun-state.mjs` or\n`node .claude\u002Fskills\u002Fblueprint-doctor\u002Fscripts\u002Frun-state.mjs`, with `start --command\n\u003Cskill> --summary \"\u003Ctruthful summary>\" --boundary\n\u003Cread-only|reviewed|local-only>`. Use `update` at milestones, with `--status\nblocked --resume \u003Ccommand>` for a blocker, and `finish --status ready|completed`\nat the end; `--help` lists every option and `blueprint\u002Fworkflow.md` has the full\nactivity contract. A missing or failing helper is a warning, not a blocker.\nCombine it with the first preflight tool batch when supported.\n\nIn one preflight batch, read:\n\n- `blueprint\u002Fconfig.json`\n- `blueprint\u002Fcontext\u002Fcurrent-feature.md`\n- the current branch, status, and recent relevant log\n- the Commands section already loaded from `AGENTS.md`\n\nStop for `\u002Fblueprint-doctor` on invalid config. Stop for `\u002Ffeature` or `\u002Ffix`\nwhen the active spec is empty. Preserve unrelated worktree changes.\n\nDo not reread the full project overview, coding standards, interaction guide,\nhistory, findings, or review ledger before coding. The approved spec already\ncontains the product contract and applicable conventions. Read one targeted\nsection only when the spec explicitly depends on a missing detail. Read findings\nand review state once at the final handoff.\n\nInspect the implementation surface in one targeted, batched read before editing.\nUse one additional read batch only when an exact dependency remains unknown and\nblocks the next change. Do not list or survey the repository, inspect unrelated\nexamples, or run scratch environment probes. Follow the declared runtime and\nexisting target-area patterns, then let the narrow check expose incompatibilities.\n\nIf the spec says `Type: Rollback`, it predates the removal of the rollback\nworkflow. Stop without changing product files and suggest removing the feature\nthrough `\u002Ffix` instead.\n\nResolve the work branch before editing. Use the exact `**Branch:**` value in the\nspec when present. For an older spec without it, combine the configured prefix\nfor its type with the work title: lowercase ASCII letters and digits, replace\neach run of other characters with one hyphen, and trim edge hyphens. Feature\ntitles come from the named build-plan item; fix titles come from their spec\nheading. Stop if the type or title is ambiguous. Create or\nswitch to that exact branch and never implement on the default branch. On\nresume, start at the first unchecked build step and use git status plus the\nchecked boxes to distinguish finished work from unfinished work.\n\n## Build loop\n\nFollow build steps in order. Build only what the spec says. If a step requires an\nunresolved product decision, unsafe action, missing prerequisite, or material\nscope expansion, stop and revise the spec instead of improvising.\n\nAdd an abstraction, dependency, service, configuration surface, compatibility\nlayer, or security mechanism only when the approved spec or an established\nrepository requirement needs it now. Prefer existing code, the standard library,\nnative platform features, and installed dependencies. If the simplest complete\nimplementation conflicts with the spec, stop and revise the spec instead of\nsilently building a larger design.\n\nFor each step:\n\n1. Make the smallest coherent change that satisfies its `Done when`.\n2. Add focused tests with logic when a test runner is configured. Never install\n   a runner or runtime dependency unless the spec authorizes it.\n3. Run the narrowest useful check while iterating. Do not run the full Verify\n   command after every step when `workflow.stepReview` is `feature`; run it once\n   after all steps. Run Verify earlier only when the step explicitly requires it\n   or later work cannot proceed without it.\n4. Self-review the diff for contract coverage, authorization and tenant scope,\n   error handling, accidental scope, and unrelated changes.\n5. Check the step box only after its code and focused check pass. Mark a repaired\n   finding `fixed`, never `closed`.\n\nWith `verification.logicTests: required`, any logic-bearing step stops and\npoints to `\u002Ftests` when no test runner is configured. Its focused logic tests\nmust pass before the step can be checked. With `verification.uiEvidence:\nrequired`, a UI done-when cannot pass on build output alone. Capture the\nconfigured browser evidence, or stop and ask the user to start the required\nserver when live evidence cannot run automatically.\n\nWith `workflow.stepReview: feature`, continue through passing steps and present\none final review packet, with no per-step choices. With `workflow.stepReview:\nevery`, stop after each step with the diff, a short explanation, evidence, and a\nmanual try path when one exists, then offer these choices:\n\n1. Continue to the next step.\n2. Commit a checkpoint, then continue. Offer this only when\n   `workflow.checkpointCommits` is `enabled`.\n3. Walk me through this step.\n4. Stop here.\n\nContinue only after the user picks continue or a checkpoint. A reply asking for\nchanges revises this step, then the choices return. A step walkthrough is a\nread-only tour of this step's diff: its purpose, key files and symbols, and\nnon-obvious decisions; offer the choices again afterward. Stop here leaves the\nfinished step checked so a later `\u002Fimplement` resumes at the next one.\n\nThe user's checkpoint pick is the commit approval for that step only. Before\ncommitting, run Verify, or the fallback build and tests, and stop if it fails.\nCommit the step's code, tests, and updated `current-feature.md` checkbox on the\nwork branch with a conventional message such as `feat: checkpoint \u003Cstep>`. Never\ncommit without current approval. `\u002Fcomplete` owns the final work-level commit\nand merge.\n\nDo not create a separate tool round merely to narrate a passing internal step.\nKeep the durable checkbox current and continue. Split a step when its diff is too\nlarge to review.\n\nBefore final verification, compare every In scope item and `Done when` against\nthe finished diff. For user-facing work, inspect the reachability and error\nclassification of each required state. Catch only known expected errors at a\nboundary; unexpected failures must reach the unexpected-error path. Fix any\nmissing or contradicted contract before marking the spec verified.\n\nRun a proportionality check before final verification: every new abstraction,\ndependency, service, configuration surface, compatibility layer, and security\nmechanism must trace to the approved spec or an established repository\nrequirement. Remove speculative machinery this work added without weakening real\ntrust-boundary validation, data-loss prevention, accessibility, or configured\nverification.\n\n## Verification\n\nAfter all steps pass, run the project's final automated gate once. If\n`AGENTS.md` declares a `Verify` command, run that exact command. Otherwise run\nthe fallback build and tests that are actually declared. Never claim a check\npassed without its output.\n\nApply configured regular gates:\n\n- Audit and independent review follow `qualityGates.regular`; `when-sensitive`\n  selects them for authentication, authorization, payments, secrets, personal\n  or user data, migrations, destructive operations, external side effects,\n  security boundaries, or unusually broad changes.\n- Check runs for `always`, for behavioral work under `when-behavioral`, or when\n  explicitly requested.\n- `\u002Fcheck guide` follows `qualityGates.regular.tryGuide`: it runs for `always`,\n  for user-facing work under `when-user-facing`, for work with a safe meaningful\n  direct observation under `when-testable`, or when explicitly requested.\n\nAfter final Verify and required Check pass, apply the selected try-guide policy.\nFor `when-testable`, follow only `..\u002Fcheck\u002Freference\u002Ftry-it-now.md` for its\nconservative classification; the full guide is not needed. Other selected\npolicies run `\u002Fcheck guide`. Keep the generated `## Try it now` section for the\nfinal packet. When no safe, meaningful direct observation exists, record that\nthe gate was skipped and do not fabricate a section. Guide output never counts\nas verification or user acceptance.\n\nDo not start a dev server. When a required runtime check needs one, ask the user\nto start it. Build output does not prove visual, persisted-data, authenticated,\nor end-to-end behavior.\n\nIf Verify or a required gate fails, repair only in-scope defects, rerun the\nnarrow failing check, then rerun the final gate. Stop on repeated failure,\nmissing infrastructure, or a product decision.\n\nAfter final Verify and required Check pass, set the active spec to `verified`\nwith every completed box checked. Then resolve independent review before the\nfinal packet:\n\n1. If `qualityGates.regular.independentReview` does not select review and no\n   request already exists, proceed directly to the final review packet.\n2. Otherwise show the exact product\u002Ftest checkpoint candidate and verified spec.\n   Include the spec when tracked; an intentionally ignored spec uses Audit's\n   local `Spec snapshot` contract without changing visibility. Obtain explicit\n   commit approval when the exact checkpoint does not already exist, then create\n   or use it. Configuration, including `review.independentExecution: \"automatic\"`,\n   never grants commit permission, even when normal checkpoint commits are disabled.\n   A local-spec-only revision may reuse the same approved product HEAD after\n   normal spec and verification gates, with a new snapshot\u002Frequest and full\n   fresh review. Do not create an empty commit for ignored spec changes.\n3. Follow `\u002Faudit independent current` to prepare or reuse the request and record\n   `Requested execution`. For requested `automatic`, start and wait for the\n   generic isolated current-runtime child instructed from the project-local\n   Audit skill, then validate the receipt. For requested `manual`, or when\n   automatic capability or access to the same local spec\u002Fsnapshot is unavailable,\n   preserve the request and stop with the manual fresh-session handoff. Treat an\n   existing request without `Requested execution` as legacy manual-only:\n   never add execution fields or run a subagent against it.\n4. Continue to the final packet only with a current passing receipt whose\n   requested execution, actual execution, and reviewer context form an allowed\n   pairing. Never self-review or silently skip a selected gate.\n\n## Final handoff\n\nRead `blueprint\u002Fcontext\u002Ffindings.md` and `blueprint\u002Fcontext\u002Freview.md` once.\nOpen or fixed P0\u002FP1 findings block `\u002Fcomplete`. Repair an open blocker as a new\nspec checklist step, mark it fixed after its check passes, then send it back to\n`\u002Faudit` for closure. Only the user can accept a finding.\n\nWhen all steps and required gates pass:\n\n- Update activity to `ready` with `\u002Fcomplete` as the resume command.\n- Present the branch, changes grouped by area, exact checks run, findings and\n  independent-review state, known risks, configured gate outcomes, and\n  `\u002Fcomplete` as the next action.\n- Include the generated `## Try it now` section verbatim when guide mode produced\n  one. Do not replace it with a vague summary or a pointer. When the selected\n  policy found no safe direct path, omit the section and state that outcome with\n  the gate results.\n\nAfter the final packet, always offer these choices:\n\n1. Walk me through the implementation.\n2. Request changes.\n3. Continue to the exact next workflow command.\n\nThe final walkthrough is available with either `workflow.stepReview` value and\nregardless of `workflow.checkpointCommits`. It is a read-only code tour, not the\nmanual product-review path produced by `\u002Fcheck guide`, and it is not verification.\n\nWhen the user chooses the walkthrough, begin with a short map of the completed\nfeature, then follow the spec's build steps. For each step, explain its purpose,\nkey files and symbols, important data or control flow, and non-obvious decisions.\nUse file and line links when the client supports them. Do not narrate every line\nor reload broad project context. End by offering a focused deep dive into one\nnamed area. If the feature spans too many distinct areas for one useful pass,\nname the sections first and let the user choose where to begin. Remain read-only\nunless the user separately requests changes.\n\nThe only commits this skill may make are an approved step checkpoint from the\nBuild loop and exactly one immutable independent-review checkpoint after showing\nits exact candidate and receiving current explicit commit approval.\nConfiguration never supplies either approval. Never create any other commit,\nincluding a product or work-level commit, and never merge, push, deploy,\npublish, or start unrelated work from this skill.\n","# implement - build the approved spec\n\n**Context reuse:** Reuse any required file already loaded in project instructions or the current session. Read it again only if absent, changed, or exact current bytes or line references are needed.\n\nThe approved `blueprint\u002Fcontext\u002Fcurrent-feature.md` is the authoritative feature\npacket. Implement it and stop before the work-level commit or merge.\n\n## Start\n\n**First action:** Before project inspection, preflight, or any other tool call,\npublish the `implement` activity as `running` when `blueprint\u002F.state\u002F` exists.\nFrom the project root, run the first helper that exists,\n`node .agents\u002Fskills\u002Fblueprint-doctor\u002Fscripts\u002Frun-state.mjs` or\n`node .claude\u002Fskills\u002Fblueprint-doctor\u002Fscripts\u002Frun-state.mjs`, with `start --command\n\u003Cskill> --summary \"\u003Ctruthful summary>\" --boundary\n\u003Cread-only|reviewed|local-only>`. Use `update` at milestones, with `--status\nblocked --resume \u003Ccommand>` for a blocker, and `finish --status ready|completed`\nat the end; `--help` lists every option and `blueprint\u002Fworkflow.md` has the full\nactivity contract. A missing or failing helper is a warning, not a blocker.\nCombine it with the first preflight tool batch when supported.\n\nIn one preflight batch, read:\n\n- `blueprint\u002Fconfig.json`\n- `blueprint\u002Fcontext\u002Fcurrent-feature.md`\n- the current branch, status, and recent relevant log\n- the Commands section already loaded from `AGENTS.md`\n\nStop for `\u002Fblueprint-doctor` on invalid config. Stop for `\u002Ffeature` or `\u002Ffix`\nwhen the active spec is empty. Preserve unrelated worktree changes.\n\nDo not reread the full project overview, coding standards, interaction guide,\nhistory, findings, or review ledger before coding. The approved spec already\ncontains the product contract and applicable conventions. Read one targeted\nsection only when the spec explicitly depends on a missing detail. Read findings\nand review state once at the final handoff.\n\nInspect the implementation surface in one targeted, batched read before editing.\nUse one additional read batch only when an exact dependency remains unknown and\nblocks the next change. Do not list or survey the repository, inspect unrelated\nexamples, or run scratch environment probes. Follow the declared runtime and\nexisting target-area patterns, then let the narrow check expose incompatibilities.\n\nIf the spec says `Type: Rollback`, it predates the removal of the rollback\nworkflow. Stop without changing product files and suggest removing the feature\nthrough `\u002Ffix` instead.\n\nResolve the work branch before editing. Use the exact `**Branch:**` value in the\nspec when present. For an older spec without it, combine the configured prefix\nfor its type with the work title: lowercase ASCII letters and digits, replace\neach run of other characters with one hyphen, and trim edge hyphens. Feature\ntitles come from the named build-plan item; fix titles come from their spec\nheading. Stop if the type or title is ambiguous. Create or\nswitch to that exact branch and never implement on the default branch. On\nresume, start at the first unchecked build step and use git status plus the\nchecked boxes to distinguish finished work from unfinished work.\n\n## Build loop\n\nFollow build steps in order. Build only what the spec says. If a step requires an\nunresolved product decision, unsafe action, missing prerequisite, or material\nscope expansion, stop and revise the spec instead of improvising.\n\nAdd an abstraction, dependency, service, configuration surface, compatibility\nlayer, or security mechanism only when the approved spec or an established\nrepository requirement needs it now. Prefer existing code, the standard library,\nnative platform features, and installed dependencies. If the simplest complete\nimplementation conflicts with the spec, stop and revise the spec instead of\nsilently building a larger design.\n\nFor each step:\n\n1. Make the smallest coherent change that satisfies its `Done when`.\n2. Add focused tests with logic when a test runner is configured. Never install\n   a runner or runtime dependency unless the spec authorizes it.\n3. Run the narrowest useful check while iterating. Do not run the full Verify\n   command after every step when `workflow.stepReview` is `feature`; run it once\n   after all steps. Run Verify earlier only when the step explicitly requires it\n   or later work cannot proceed without it.\n4. Self-review the diff for contract coverage, authorization and tenant scope,\n   error handling, accidental scope, and unrelated changes.\n5. Check the step box only after its code and focused check pass. Mark a repaired\n   finding `fixed`, never `closed`.\n\nWith `verification.logicTests: required`, any logic-bearing step stops and\npoints to `\u002Ftests` when no test runner is configured. Its focused logic tests\nmust pass before the step can be checked. With `verification.uiEvidence:\nrequired`, a UI done-when cannot pass on build output alone. Capture the\nconfigured browser evidence, or stop and ask the user to start the required\nserver when live evidence cannot run automatically.\n\nWith `workflow.stepReview: feature`, continue through passing steps and present\none final review packet, with no per-step choices. With `workflow.stepReview:\nevery`, stop after each step with the diff, a short explanation, evidence, and a\nmanual try path when one exists, then offer these choices:\n\n1. Continue to the next step.\n2. Commit a checkpoint, then continue. Offer this only when\n   `workflow.checkpointCommits` is `enabled`.\n3. Walk me through this step.\n4. Stop here.\n\nContinue only after the user picks continue or a checkpoint. A reply asking for\nchanges revises this step, then the choices return. A step walkthrough is a\nread-only tour of this step's diff: its purpose, key files and symbols, and\nnon-obvious decisions; offer the choices again afterward. Stop here leaves the\nfinished step checked so a later `\u002Fimplement` resumes at the next one.\n\nThe user's checkpoint pick is the commit approval for that step only. Before\ncommitting, run Verify, or the fallback build and tests, and stop if it fails.\nCommit the step's code, tests, and updated `current-feature.md` checkbox on the\nwork branch with a conventional message such as `feat: checkpoint \u003Cstep>`. Never\ncommit without current approval. `\u002Fcomplete` owns the final work-level commit\nand merge.\n\nDo not create a separate tool round merely to narrate a passing internal step.\nKeep the durable checkbox current and continue. Split a step when its diff is too\nlarge to review.\n\nBefore final verification, compare every In scope item and `Done when` against\nthe finished diff. For user-facing work, inspect the reachability and error\nclassification of each required state. Catch only known expected errors at a\nboundary; unexpected failures must reach the unexpected-error path. Fix any\nmissing or contradicted contract before marking the spec verified.\n\nRun a proportionality check before final verification: every new abstraction,\ndependency, service, configuration surface, compatibility layer, and security\nmechanism must trace to the approved spec or an established repository\nrequirement. Remove speculative machinery this work added without weakening real\ntrust-boundary validation, data-loss prevention, accessibility, or configured\nverification.\n\n## Verification\n\nAfter all steps pass, run the project's final automated gate once. If\n`AGENTS.md` declares a `Verify` command, run that exact command. Otherwise run\nthe fallback build and tests that are actually declared. Never claim a check\npassed without its output.\n\nApply configured regular gates:\n\n- Audit and independent review follow `qualityGates.regular`; `when-sensitive`\n  selects them for authentication, authorization, payments, secrets, personal\n  or user data, migrations, destructive operations, external side effects,\n  security boundaries, or unusually broad changes.\n- Check runs for `always`, for behavioral work under `when-behavioral`, or when\n  explicitly requested.\n- `\u002Fcheck guide` follows `qualityGates.regular.tryGuide`: it runs for `always`,\n  for user-facing work under `when-user-facing`, for work with a safe meaningful\n  direct observation under `when-testable`, or when explicitly requested.\n\nAfter final Verify and required Check pass, apply the selected try-guide policy.\nFor `when-testable`, follow only `..\u002Fcheck\u002Freference\u002Ftry-it-now.md` for its\nconservative classification; the full guide is not needed. Other selected\npolicies run `\u002Fcheck guide`. Keep the generated `## Try it now` section for the\nfinal packet. When no safe, meaningful direct observation exists, record that\nthe gate was skipped and do not fabricate a section. Guide output never counts\nas verification or user acceptance.\n\nDo not start a dev server. When a required runtime check needs one, ask the user\nto start it. Build output does not prove visual, persisted-data, authenticated,\nor end-to-end behavior.\n\nIf Verify or a required gate fails, repair only in-scope defects, rerun the\nnarrow failing check, then rerun the final gate. Stop on repeated failure,\nmissing infrastructure, or a product decision.\n\nAfter final Verify and required Check pass, set the active spec to `verified`\nwith every completed box checked. Then resolve independent review before the\nfinal packet:\n\n1. If `qualityGates.regular.independentReview` does not select review and no\n   request already exists, proceed directly to the final review packet.\n2. Otherwise show the exact product\u002Ftest checkpoint candidate and verified spec.\n   Include the spec when tracked; an intentionally ignored spec uses Audit's\n   local `Spec snapshot` contract without changing visibility. Obtain explicit\n   commit approval when the exact checkpoint does not already exist, then create\n   or use it. Configuration, including `review.independentExecution: \"automatic\"`,\n   never grants commit permission, even when normal checkpoint commits are disabled.\n   A local-spec-only revision may reuse the same approved product HEAD after\n   normal spec and verification gates, with a new snapshot\u002Frequest and full\n   fresh review. Do not create an empty commit for ignored spec changes.\n3. Follow `\u002Faudit independent current` to prepare or reuse the request and record\n   `Requested execution`. For requested `automatic`, start and wait for the\n   generic isolated current-runtime child instructed from the project-local\n   Audit skill, then validate the receipt. For requested `manual`, or when\n   automatic capability or access to the same local spec\u002Fsnapshot is unavailable,\n   preserve the request and stop with the manual fresh-session handoff. Treat an\n   existing request without `Requested execution` as legacy manual-only:\n   never add execution fields or run a subagent against it.\n4. Continue to the final packet only with a current passing receipt whose\n   requested execution, actual execution, and reviewer context form an allowed\n   pairing. Never self-review or silently skip a selected gate.\n\n## Final handoff\n\nRead `blueprint\u002Fcontext\u002Ffindings.md` and `blueprint\u002Fcontext\u002Freview.md` once.\nOpen or fixed P0\u002FP1 findings block `\u002Fcomplete`. Repair an open blocker as a new\nspec checklist step, mark it fixed after its check passes, then send it back to\n`\u002Faudit` for closure. Only the user can accept a finding.\n\nWhen all steps and required gates pass:\n\n- Update activity to `ready` with `\u002Fcomplete` as the resume command.\n- Present the branch, changes grouped by area, exact checks run, findings and\n  independent-review state, known risks, configured gate outcomes, and\n  `\u002Fcomplete` as the next action.\n- Include the generated `## Try it now` section verbatim when guide mode produced\n  one. Do not replace it with a vague summary or a pointer. When the selected\n  policy found no safe direct path, omit the section and state that outcome with\n  the gate results.\n\nAfter the final packet, always offer these choices:\n\n1. Walk me through the implementation.\n2. Request changes.\n3. Continue to the exact next workflow command.\n\nThe final walkthrough is available with either `workflow.stepReview` value and\nregardless of `workflow.checkpointCommits`. It is a read-only code tour, not the\nmanual product-review path produced by `\u002Fcheck guide`, and it is not verification.\n\nWhen the user chooses the walkthrough, begin with a short map of the completed\nfeature, then follow the spec's build steps. For each step, explain its purpose,\nkey files and symbols, important data or control flow, and non-obvious decisions.\nUse file and line links when the client supports them. Do not narrate every line\nor reload broad project context. End by offering a focused deep dive into one\nnamed area. If the feature spans too many distinct areas for one useful pass,\nname the sections first and let the user choose where to begin. Remain read-only\nunless the user separately requests changes.\n\nThe only commits this skill may make are an approved step checkpoint from the\nBuild loop and exactly one immutable independent-review checkpoint after showing\nits exact candidate and receiving current explicit commit approval.\nConfiguration never supplies either approval. Never create any other commit,\nincluding a product or work-level commit, and never merge, push, deploy,\npublish, or start unrelated work from this skill.\n",{"name":5,"description":7,"disable-model-invocation":23},true,"main","2ac6d59a4c57a9d69ce652adca1976d8ef8f182c",".claude\u002Fskills\u002Fimplement",132,"https:\u002F\u002Fgithub.com\u002Faiblueprinthq\u002Fai-blueprint","https:\u002F\u002Fai-blueprint.dev","https:\u002F\u002Fgithub.com\u002Faiblueprinthq","2026-11-14 19:46:05",{"source":8,"skillName":5,"commands":33,"prompt":43,"zipAvailable":23},[34,37,40],{"tool":35,"command":36},"npx","npx skills add aiblueprinthq\u002Fai-blueprint --skill implement",{"tool":38,"command":39},"bunx","bunx skills add aiblueprinthq\u002Fai-blueprint --skill implement",{"tool":41,"command":42},"pnpm","pnpm dlx skills add aiblueprinthq\u002Fai-blueprint --skill implement","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Faiblueprinthq\u002Fai-blueprint\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fimplement\u002FSKILL.md",[],[46,52,57,64,71,78],{"skillNo":47,"name":48,"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":49,"categories":50,"keywords":51,"publishedAt":18},"01M75Y34X26C74A66RYE49W8JY","status","https:\u002F\u002Fgithub.com\u002Faiblueprinthq\u002Fai-blueprint\u002Fblob\u002Fmain\u002Fpackages\u002Fcreate-ai-blueprint\u002Ftest\u002Ffixtures\u002Fhealthy-blueprint\u002F.claude\u002Fskills\u002Fstatus\u002FSKILL.md",[],[],{"skillNo":53,"name":48,"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":54,"categories":55,"keywords":56,"publishedAt":18},"01M75Y3456H4WZZ25634QY9KRA","https:\u002F\u002Fgithub.com\u002Faiblueprinthq\u002Fai-blueprint\u002Fblob\u002Fmain\u002Fpackages\u002Fcreate-ai-blueprint\u002Ftest\u002Ffixtures\u002Fhealthy-blueprint\u002F.agents\u002Fskills\u002Fstatus\u002FSKILL.md",[],[],{"skillNo":58,"name":59,"nameZh":6,"description":60,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":61,"categories":62,"keywords":63,"publishedAt":18},"01M75Y31WCYVFM5M4NH7D9MRMW","tests","Set up unit testing with \u002Ftests or \u002Ftests unit, or explicitly set up browser and end-to-end testing with \u002Ftests browser. Reuse or install a suitable test runner, add one example or smoke test, document commands, and verify. Use check for one-time live verification.","https:\u002F\u002Fgithub.com\u002Faiblueprinthq\u002Fai-blueprint\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Ftests\u002FSKILL.md",[],[],{"skillNo":65,"name":66,"nameZh":6,"description":67,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":68,"categories":69,"keywords":70,"publishedAt":18},"01M75Y2ZETQZEHN2V19HWZBB1R","prototype","Plan and create throwaway static HTML and CSS mockups with shared design tokens before feature implementation. Use for \u002Fprototype, screen mockups, layout exploration, themes, or deciding a project's look and feel.","https:\u002F\u002Fgithub.com\u002Faiblueprinthq\u002Fai-blueprint\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fprototype\u002FSKILL.md",[],[],{"skillNo":72,"name":73,"nameZh":6,"description":74,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":75,"categories":76,"keywords":77,"publishedAt":18},"01M75Y2WY81F9N40G2RADX2FKH","overview","Validate and normalize project-plan.md and build-plan.md, then generate the durable project-overview.md used by agents. Use for \u002Foverview, plan cleanup, generating the first overview, or refreshing context after either plan changes.","https:\u002F\u002Fgithub.com\u002Faiblueprinthq\u002Fai-blueprint\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Foverview\u002FSKILL.md",[],[],{"skillNo":79,"name":80,"nameZh":6,"description":81,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":82,"categories":83,"keywords":84,"publishedAt":18},"01M75Y2SNF3FZJP3NHRT0B5MQE","onboard","Onboard a fresh or early scaffold after Blueprint is overlaid by tuning commands, standards, adapters, visibility, and context loading. Use for \u002Fonboard, fresh installation setup, or what to do after installing Blueprint. Use adopt for an established app.","https:\u002F\u002Fgithub.com\u002Faiblueprinthq\u002Fai-blueprint\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fonboard\u002FSKILL.md",[],[]]