[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M79DZDAXFEEKH0VDPAVWED56":3,"skill-same:01M79DZDAXFEEKH0VDPAVWED56":47},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":24,"commitSha":25,"skillDirectory":26,"forksCount":14,"repositoryHtmlUrl":27,"repositoryHomepageUrl":28,"ownerHtmlUrl":29,"ownerBlogUrl":30,"ownerDescription":31,"updatedAt":32,"install":33,"resources":46},"01M79DZDAXFEEKH0VDPAVWED56","plan_and_execute_feature",null,"Use when delivering a feature through explicit planning, phased execution, validation, and governed handoff — or when implementing\u002Ffixing already-scoped engineering work via the execute_only mode.","marcosdh1987\u002Fseparator-control-challenge","marcosdh1987","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F31476977?v=4","01M6C0RHZYKQDC51CRAMD70QPA","Marcos Soto","01M79DY9HHXHREQ9Z7QX2R97YG",0,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Fmarcosdh1987\u002Fseparator-control-challenge\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fplan_and_execute_feature\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: plan_and_execute_feature\ndescription: Use when delivering a feature through explicit planning, phased execution, validation, and governed handoff — or when implementing\u002Ffixing already-scoped engineering work via the execute_only mode.\n---\n\n# Skill: plan_and_execute_feature\n\n## Purpose\n\nDeliver a feature through explicit orchestration phases with architecture-safe execution.\n\n## Mode\n\n- `mode: full` (default) — run all five phases below, from planning to validation.\n- `mode: execute_only` — a plan is already approved; skip Phases 1-2 and enter directly\n  at Phase 4 (Execution), then Phase 5 (Validation). Use this for implementing a feature,\n  fixing a bug, or executing scoped engineering work through governed orchestration\n  instead of ad-hoc generation.\n- `mode: local_model_32k` — for small self-hosted models (Qwen 3.x \u002F 3.6, 27B-35B @ 32k)\n  driving spikes \u002F vibe coding via Copilot \u002F OpenCode. Never deliver in one response: emit a\n  small single-file backlog and execute one step per turn. See the weak-model section below\n  and `LOCAL_AGENT.md`.\n\n## Required Input\n\n- Feature request.\n- Business outcome.\n- Affected modules and boundaries.\n- Non-functional requirements (performance, security, observability).\n- Test expectations.\n- Optional `mode` (`full` | `execute_only`). For `execute_only`: the approved plan,\n  scope constraints (files\u002Fmodules in and out of scope), and acceptance criteria.\n\n## Output Format\n\n- Phase report (1 to 5).\n- Approved implementation plan.\n- Execution backlog: the plan as an ordered list of steps where each step is\n  independently verifiable (compiles \u002F a test passes \u002F it runs), names the one or\n  few files it touches, and is annotated with the target tier\n  (`planner` \u002F `executor` \u002F `fast`). Size steps to that tier — see\n  `docs\u002Ftask-sizing.md`.\n- Selected skills map.\n- Code\u002Ftest\u002Fdoc change summary.\n- Validation checklist with pass\u002Ffail evidence.\n\n## Phases\n\n### Phase 1 - Planning\n\n- Clarify scope, assumptions, and acceptance criteria.\n- Build a step-by-step implementation plan.\n\n### Phase 2 - Architecture Validation\n\n- Validate intended changes against:\n  - `.github\u002Farchitecture.md`\n  - `.github\u002Fdomain-boundaries.md`\n- Adjust plan to preserve dependency direction and module placement.\n\n### Phase 3 - Skill Selection\n\n- Select internal operational skills needed for execution.\n- Define invocation order and dependencies between skills.\n\n### Phase 4 - Execution\n\n- Before editing files, publish a short visible plan: scope, files likely to change,\n  tests to add or run, risks\u002Fassumptions. (In IDE agents like Antigravity that lack a\n  separate plan surface, state it inline before the first edit.)\n- Implement changes following selected skills and plan order.\n- Execute in small checkpoints; update the plan when it changes.\n- Keep modifications scoped and documented.\n\n### Phase 5 - Validation\n\n- Validate against `.github\u002Fautomation.md` and `.github\u002Forchestration.md`.\n- Review diffs, run required checks, and produce final summary.\n\n## Small-context \u002F weak-model mode\n\nWhen the executor is a weak or self-hosted model (small context window, weaker\ninstruction-following), the default full flow can loop without converging. Adapt:\n\n- Treat the work as a `spike` first (flat structure, runnable milestone) and\n  promote to clean architecture only once it runs — see `docs\u002Ftask-sizing.md`.\n- Execute **one file (or one function) per turn**; run or test it before the next.\n- Make the first deliverable the smallest thing that **runs**, not the most\n  correct or complete one; add one capability per loop.\n- Keep the same model for planning and execution, or a planner only slightly\n  stronger, so backlog steps stay sized to what the executor can chew.\n- If a step cannot be stated as \"produce\u002Fmodify this one file and verify it\", it\n  is too big — split it before executing.\n\nUnder Copilot agent mode specifically (`mode: local_model_32k`), add an output discipline —\nweak models otherwise try to regenerate whole files and fail with `Response too long`:\n\n- **Edit, never rewrite.** Change only the wrong lines; never regenerate a whole file.\n- **One root cause per turn.** If asked to \"fix everything\", fix the single most critical\n  cause, list the rest deferred (one line each), and stop.\n- **Small output, fixed shape:** `Target file:` \u002F `Expected change:` \u002F `Validation:`, then the\n  edit, then a one-line result. No analysis or summary prose (it accelerates compaction).\n- This is enforced mechanically by a gateway `max_tokens` cap — see `LOCAL_AGENT.md` and\n  [`..\u002F..\u002Fdocs\u002Flocal-model-runtime-config.md`](..\u002F..\u002Fdocs\u002Flocal-model-runtime-config.md).\n\n## Execution Rules\n\n1. Internal skills have precedence over external synced skills.\n2. In `mode: full`, no direct implementation without completing Phase 1 and Phase 2.\n3. Prefer minimal file churn and strict scope boundaries.\n4. Include tests and `docs\u002F` updates when behavior changes.\n5. Escalate blockers with explicit assumptions and next actions.\n6. Report open risks and deferred items explicitly.\n","# Skill: plan_and_execute_feature\n\n## Purpose\n\nDeliver a feature through explicit orchestration phases with architecture-safe execution.\n\n## Mode\n\n- `mode: full` (default) — run all five phases below, from planning to validation.\n- `mode: execute_only` — a plan is already approved; skip Phases 1-2 and enter directly\n  at Phase 4 (Execution), then Phase 5 (Validation). Use this for implementing a feature,\n  fixing a bug, or executing scoped engineering work through governed orchestration\n  instead of ad-hoc generation.\n- `mode: local_model_32k` — for small self-hosted models (Qwen 3.x \u002F 3.6, 27B-35B @ 32k)\n  driving spikes \u002F vibe coding via Copilot \u002F OpenCode. Never deliver in one response: emit a\n  small single-file backlog and execute one step per turn. See the weak-model section below\n  and `LOCAL_AGENT.md`.\n\n## Required Input\n\n- Feature request.\n- Business outcome.\n- Affected modules and boundaries.\n- Non-functional requirements (performance, security, observability).\n- Test expectations.\n- Optional `mode` (`full` | `execute_only`). For `execute_only`: the approved plan,\n  scope constraints (files\u002Fmodules in and out of scope), and acceptance criteria.\n\n## Output Format\n\n- Phase report (1 to 5).\n- Approved implementation plan.\n- Execution backlog: the plan as an ordered list of steps where each step is\n  independently verifiable (compiles \u002F a test passes \u002F it runs), names the one or\n  few files it touches, and is annotated with the target tier\n  (`planner` \u002F `executor` \u002F `fast`). Size steps to that tier — see\n  `docs\u002Ftask-sizing.md`.\n- Selected skills map.\n- Code\u002Ftest\u002Fdoc change summary.\n- Validation checklist with pass\u002Ffail evidence.\n\n## Phases\n\n### Phase 1 - Planning\n\n- Clarify scope, assumptions, and acceptance criteria.\n- Build a step-by-step implementation plan.\n\n### Phase 2 - Architecture Validation\n\n- Validate intended changes against:\n  - `.github\u002Farchitecture.md`\n  - `.github\u002Fdomain-boundaries.md`\n- Adjust plan to preserve dependency direction and module placement.\n\n### Phase 3 - Skill Selection\n\n- Select internal operational skills needed for execution.\n- Define invocation order and dependencies between skills.\n\n### Phase 4 - Execution\n\n- Before editing files, publish a short visible plan: scope, files likely to change,\n  tests to add or run, risks\u002Fassumptions. (In IDE agents like Antigravity that lack a\n  separate plan surface, state it inline before the first edit.)\n- Implement changes following selected skills and plan order.\n- Execute in small checkpoints; update the plan when it changes.\n- Keep modifications scoped and documented.\n\n### Phase 5 - Validation\n\n- Validate against `.github\u002Fautomation.md` and `.github\u002Forchestration.md`.\n- Review diffs, run required checks, and produce final summary.\n\n## Small-context \u002F weak-model mode\n\nWhen the executor is a weak or self-hosted model (small context window, weaker\ninstruction-following), the default full flow can loop without converging. Adapt:\n\n- Treat the work as a `spike` first (flat structure, runnable milestone) and\n  promote to clean architecture only once it runs — see `docs\u002Ftask-sizing.md`.\n- Execute **one file (or one function) per turn**; run or test it before the next.\n- Make the first deliverable the smallest thing that **runs**, not the most\n  correct or complete one; add one capability per loop.\n- Keep the same model for planning and execution, or a planner only slightly\n  stronger, so backlog steps stay sized to what the executor can chew.\n- If a step cannot be stated as \"produce\u002Fmodify this one file and verify it\", it\n  is too big — split it before executing.\n\nUnder Copilot agent mode specifically (`mode: local_model_32k`), add an output discipline —\nweak models otherwise try to regenerate whole files and fail with `Response too long`:\n\n- **Edit, never rewrite.** Change only the wrong lines; never regenerate a whole file.\n- **One root cause per turn.** If asked to \"fix everything\", fix the single most critical\n  cause, list the rest deferred (one line each), and stop.\n- **Small output, fixed shape:** `Target file:` \u002F `Expected change:` \u002F `Validation:`, then the\n  edit, then a one-line result. No analysis or summary prose (it accelerates compaction).\n- This is enforced mechanically by a gateway `max_tokens` cap — see `LOCAL_AGENT.md` and\n  [`..\u002F..\u002Fdocs\u002Flocal-model-runtime-config.md`](..\u002F..\u002Fdocs\u002Flocal-model-runtime-config.md).\n\n## Execution Rules\n\n1. Internal skills have precedence over external synced skills.\n2. In `mode: full`, no direct implementation without completing Phase 1 and Phase 2.\n3. Prefer minimal file churn and strict scope boundaries.\n4. Include tests and `docs\u002F` updates when behavior changes.\n5. Escalate blockers with explicit assumptions and next actions.\n6. Report open risks and deferred items explicitly.\n",{"name":5,"description":7},"main","73b8401646484c5eb5e5b45a63607f118bc58d14",".agents\u002Fskills\u002Fplan_and_execute_feature","https:\u002F\u002Fgithub.com\u002Fmarcosdh1987\u002Fseparator-control-challenge","https:\u002F\u002Fseparator-control-challenge.vercel.app","https:\u002F\u002Fgithub.com\u002Fmarcosdh1987","https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Fmarcosesoto\u002F","I'm a System Engineer, a big fan of technology and travels ","2026-09-13 21:51:09",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":45},[35,38,41],{"tool":36,"command":37},"npx","npx skills add marcosdh1987\u002Fseparator-control-challenge --skill plan_and_execute_feature",{"tool":39,"command":40},"bunx","bunx skills add marcosdh1987\u002Fseparator-control-challenge --skill plan_and_execute_feature",{"tool":42,"command":43},"pnpm","pnpm dlx skills add marcosdh1987\u002Fseparator-control-challenge --skill plan_and_execute_feature","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fmarcosdh1987\u002Fseparator-control-challenge\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fplan_and_execute_feature\u002FSKILL.md",true,[],[48,54,60,66,72,78],{"skillNo":49,"name":50,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":51,"categories":52,"keywords":53,"publishedAt":19},"01M79E5D38BN8PQ3WJ8YFP4CNT","writing-plans","https:\u002F\u002Fgithub.com\u002Fmarcosdh1987\u002Fseparator-control-challenge\u002Fblob\u002Fmain\u002F.opencode\u002Fskills\u002Fwriting-plans\u002FSKILL.md",[],[],{"skillNo":55,"name":56,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":57,"categories":58,"keywords":59,"publishedAt":19},"01M79E5C05HSKKYD0SETH9G605","writing-clearly-and-concisely","https:\u002F\u002Fgithub.com\u002Fmarcosdh1987\u002Fseparator-control-challenge\u002Fblob\u002Fmain\u002F.opencode\u002Fskills\u002Fwriting-clearly-and-concisely\u002FSKILL.md",[],[],{"skillNo":61,"name":62,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":63,"categories":64,"keywords":65,"publishedAt":19},"01M79E5ANYKVWFZES04JS478AK","verify_changes","https:\u002F\u002Fgithub.com\u002Fmarcosdh1987\u002Fseparator-control-challenge\u002Fblob\u002Fmain\u002F.opencode\u002Fskills\u002Fverify_changes\u002FSKILL.md",[],[],{"skillNo":67,"name":68,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":69,"categories":70,"keywords":71,"publishedAt":19},"01M79E57Q0NQ4TMW9H5RPPKVZF","validate_module_structure","https:\u002F\u002Fgithub.com\u002Fmarcosdh1987\u002Fseparator-control-challenge\u002Fblob\u002Fmain\u002F.opencode\u002Fskills\u002Fvalidate_module_structure\u002FSKILL.md",[],[],{"skillNo":73,"name":74,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":75,"categories":76,"keywords":77,"publishedAt":19},"01M79E56SP5PAWKKZ8XGMXGCP6","using-git-worktrees","https:\u002F\u002Fgithub.com\u002Fmarcosdh1987\u002Fseparator-control-challenge\u002Fblob\u002Fmain\u002F.opencode\u002Fskills\u002Fusing-git-worktrees\u002FSKILL.md",[],[],{"skillNo":79,"name":80,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":81,"categories":82,"keywords":83,"publishedAt":19},"01M79E55WJ4ECJEW1B5D9KZTN7","ui-ux-pro-max","https:\u002F\u002Fgithub.com\u002Fmarcosdh1987\u002Fseparator-control-challenge\u002Fblob\u002Fmain\u002F.opencode\u002Fskills\u002Fui-ux-pro-max\u002FSKILL.md",[],[]]