[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7C37GPHK1EWP6SDRJ9CVHBA":3,"skill-same:01M7C37GPHK1EWP6SDRJ9CVHBA":75},{"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":39,"commitSha":40,"skillDirectory":41,"forksCount":42,"repositoryHtmlUrl":43,"repositoryHomepageUrl":6,"ownerHtmlUrl":44,"ownerBlogUrl":6,"ownerDescription":45,"updatedAt":46,"install":47,"resources":60},"01M7C37GPHK1EWP6SDRJ9CVHBA","tech-feasibility",null,"Validates whether a project is technically feasible with what already exists: interviews the developer for the approach they have in mind, settles what the documentation can settle, and spikes only what reading cannot. Use when asked whether an idea is viable, whether X can be built with Y, what already solves this, or to run a spike or proof of concept.","leolicona\u002Fagents-skills","leolicona","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F84453410?v=4","01M7C37BCCMA81W2N0JSC7D0WB","Leo Licona ","01M7C37BD5R3J4Z44CSDBW7KAJ",1,"MIT","https:\u002F\u002Fgithub.com\u002Fleolicona\u002Fagents-skills\u002Fblob\u002Fmain\u002Fskills\u002Ftech-feasibility\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: tech-feasibility\ndescription: >-\n  Validates whether a project is technically feasible with what already\n  exists: interviews the developer for the approach they have in mind, settles\n  what the documentation can settle, and spikes only what reading cannot. Use\n  when asked whether an idea is viable, whether X can be built with Y, what\n  already solves this, or to run a spike or proof of concept.\ntags: [feasibility, research, spike, build-vs-buy, spec-driven-development]\nallowed-tools: [Read, Write, Edit, Glob, Grep, Bash, WebSearch, WebFetch]\n---\n\n# Technical feasibility\n\n## When to use this skill\n\nThe project rests on a technical bet: can this be built with what already\nexists, inside the budget, by this developer.\n\nOf the four product risks — value, usability, feasibility, business viability\n— this covers **feasibility** and the cost half of **viability**. Value and\nusability need users, not code: never let a green verdict pass as product\nvalidation.\n\nContext: one developer building product with AI agents, and that developer\ndecides. Their experience and the project's budget are pass\u002Ffail criteria\nhere, not background.\n\n## Three stages, cheapest first\n\nRun them in order. The cheapest test that could change the decision always\ngoes first, and most approaches die or survive before a line of code is\nwritten.\n\n**A. Interview** — what is the bet, and what would break it.\n**B. Desk research** — settle everything the documentation can settle.\n**C. Spike** — measure only what is left.\n\n## Stage A — interview\n\nRead `projectbrief.md` first for the appetite and the run-cost ceiling, and\nthe constitution (`memory\u002Fconstitution.md` or `.specify\u002Fmemory\u002Fconstitution.md`)\nfor constraints already decided. Then two rounds, three questions each.\n\n*Round 1 — the bet:*\n\n1. What will you build it with, what draws you to that, and what did you\n   consider and drop?\n2. Which of those pieces are boring for you — already in production — and\n   which are new?\n3. What would make you abandon this approach?\n\n*Round 2 — the scenario, which becomes the pass criterion:*\n\n4. Describe the moment of highest stress: what triggers it, how often, and\n   under what conditions.\n5. What must happen then, and with what number and threshold do we measure\n   it?\n6. When it fails, what should happen — and do you know how these pieces fail?\n\nThen write the hypothesis in one line, and under it the assumptions it rests\non, ranked by what sinks the approach if false. **Test the riskiest, never\nthe comfortable one**: a green result on a comfortable assumption is a dead\nproject with good paperwork.\n\nRisk concentrates in the new pieces — the developer already knows the failure\nmodes of the boring ones. An MVP affords about one new technology; if\nquestion 2 turns up three, say so now rather than after the spike.\n\n## Stage B — desk research\n\nFor each assumption, try to settle it by reading, before touching code. Name\neach capability the MVP needs in vendor-neutral terms — \"receive and verify\ninbound webhooks\", not \"use Workers\" — and ask what already solves it whole\nbefore considering building it.\n\nCheck, in this order:\n\n- **Hard limits**: rate limits, payload sizes, timeouts, message windows,\n  quotas, regions. These kill designs late; find them first.\n- **Price** at MVP scale and at 10x, against the ceiling from the brief.\n- **Runtime fit**: does it run where you already deploy? Many SDKs assume\n  APIs an edge runtime does not have.\n- **Is it alive**: last release, open issues, whether the vendor still sells\n  it.\n- **Known failure modes**, and the exit cost if it has to be swapped.\n- **Compliance**, when the capability touches personal data.\n\nEvery assumption leaves this stage in one of three states:\n\n| State | Meaning | What follows |\n| --- | --- | --- |\n| **Refuted** | the documentation says it cannot work | change the approach now — the cheapest good news you will get |\n| **Confirmed on paper** | promised by the vendor, plausible | proceed, with the residual risk written down |\n| **Unresolved** | reading cannot answer it | goes to stage C |\n\n**Documentation can kill an approach outright, but never certify one.** It\nstates what is promised, not what happens under load, cold, with your\npayloads. So *confirmed on paper* is not enough for the riskiest assumption:\nthat one still gets measured.\n\nRules for this stage:\n\n- Verify against the vendor's own documentation, never from memory. A blog\n  post is a lead, not evidence.\n- Record the URL and the date checked for every limit and price. An\n  unverified number is an open question, not a fact.\n- Record every rejected alternative in one line. Half the value of this stage\n  is what you ruled out, and it is where the fallback comes from.\n- Default to adopting what exists. Build only the differentiator — the part\n  that makes the product yours.\n\n## Stage C — spike\n\nOnly for **unresolved** assumptions, and only after stage B: never spike what\nyou did not first try to settle by reading.\n\nRead `references\u002Fspike.md` before running one. It covers the shape of the\ntest, the spike card you get approved before writing code, the spending gate,\nthe time box, the three verdicts and the cleanup.\n\nIf stage B leaves nothing unresolved, there is no spike: declare it feasible\non paper and name the residual risk.\n\n## Output\n\nWrite `feasibility.md` from `references\u002Ffeasibility-template.md`, and one\nrecord per experiment under `spikes\u002F`. Report the path, the verdict and what\nstayed open.\n\n## Rules\n\n- **Never invent.** Anything unsettled becomes\n  `[NEEDS CLARIFICATION: what is missing - optionA\u002FoptionB?]`.\n- **Never decide for the developer.** Deliver findings, verdict and a\n  recommendation; the decision is theirs.\n- **Stay inside the appetite.** Research and spikes are spent from it, not\n  added to it.\n- **One page plus the tables.** Depth goes into the spike records.\n- Never edit `specs\u002F`, the constitution or `.specify\u002F`.\n\n## Spec-driven development\n\nThis is Phase 0 research: the verdicts and limits fill the **Technical\nContext** that `\u002Fspeckit.plan` needs (language, dependencies, storage,\nplatform, constraints, scale). Hand `feasibility.md` over and let the command\nwrite `plan.md` and `research.md`.\n","# Technical feasibility\n\n## When to use this skill\n\nThe project rests on a technical bet: can this be built with what already\nexists, inside the budget, by this developer.\n\nOf the four product risks — value, usability, feasibility, business viability\n— this covers **feasibility** and the cost half of **viability**. Value and\nusability need users, not code: never let a green verdict pass as product\nvalidation.\n\nContext: one developer building product with AI agents, and that developer\ndecides. Their experience and the project's budget are pass\u002Ffail criteria\nhere, not background.\n\n## Three stages, cheapest first\n\nRun them in order. The cheapest test that could change the decision always\ngoes first, and most approaches die or survive before a line of code is\nwritten.\n\n**A. Interview** — what is the bet, and what would break it.\n**B. Desk research** — settle everything the documentation can settle.\n**C. Spike** — measure only what is left.\n\n## Stage A — interview\n\nRead `projectbrief.md` first for the appetite and the run-cost ceiling, and\nthe constitution (`memory\u002Fconstitution.md` or `.specify\u002Fmemory\u002Fconstitution.md`)\nfor constraints already decided. Then two rounds, three questions each.\n\n*Round 1 — the bet:*\n\n1. What will you build it with, what draws you to that, and what did you\n   consider and drop?\n2. Which of those pieces are boring for you — already in production — and\n   which are new?\n3. What would make you abandon this approach?\n\n*Round 2 — the scenario, which becomes the pass criterion:*\n\n4. Describe the moment of highest stress: what triggers it, how often, and\n   under what conditions.\n5. What must happen then, and with what number and threshold do we measure\n   it?\n6. When it fails, what should happen — and do you know how these pieces fail?\n\nThen write the hypothesis in one line, and under it the assumptions it rests\non, ranked by what sinks the approach if false. **Test the riskiest, never\nthe comfortable one**: a green result on a comfortable assumption is a dead\nproject with good paperwork.\n\nRisk concentrates in the new pieces — the developer already knows the failure\nmodes of the boring ones. An MVP affords about one new technology; if\nquestion 2 turns up three, say so now rather than after the spike.\n\n## Stage B — desk research\n\nFor each assumption, try to settle it by reading, before touching code. Name\neach capability the MVP needs in vendor-neutral terms — \"receive and verify\ninbound webhooks\", not \"use Workers\" — and ask what already solves it whole\nbefore considering building it.\n\nCheck, in this order:\n\n- **Hard limits**: rate limits, payload sizes, timeouts, message windows,\n  quotas, regions. These kill designs late; find them first.\n- **Price** at MVP scale and at 10x, against the ceiling from the brief.\n- **Runtime fit**: does it run where you already deploy? Many SDKs assume\n  APIs an edge runtime does not have.\n- **Is it alive**: last release, open issues, whether the vendor still sells\n  it.\n- **Known failure modes**, and the exit cost if it has to be swapped.\n- **Compliance**, when the capability touches personal data.\n\nEvery assumption leaves this stage in one of three states:\n\n| State | Meaning | What follows |\n| --- | --- | --- |\n| **Refuted** | the documentation says it cannot work | change the approach now — the cheapest good news you will get |\n| **Confirmed on paper** | promised by the vendor, plausible | proceed, with the residual risk written down |\n| **Unresolved** | reading cannot answer it | goes to stage C |\n\n**Documentation can kill an approach outright, but never certify one.** It\nstates what is promised, not what happens under load, cold, with your\npayloads. So *confirmed on paper* is not enough for the riskiest assumption:\nthat one still gets measured.\n\nRules for this stage:\n\n- Verify against the vendor's own documentation, never from memory. A blog\n  post is a lead, not evidence.\n- Record the URL and the date checked for every limit and price. An\n  unverified number is an open question, not a fact.\n- Record every rejected alternative in one line. Half the value of this stage\n  is what you ruled out, and it is where the fallback comes from.\n- Default to adopting what exists. Build only the differentiator — the part\n  that makes the product yours.\n\n## Stage C — spike\n\nOnly for **unresolved** assumptions, and only after stage B: never spike what\nyou did not first try to settle by reading.\n\nRead `references\u002Fspike.md` before running one. It covers the shape of the\ntest, the spike card you get approved before writing code, the spending gate,\nthe time box, the three verdicts and the cleanup.\n\nIf stage B leaves nothing unresolved, there is no spike: declare it feasible\non paper and name the residual risk.\n\n## Output\n\nWrite `feasibility.md` from `references\u002Ffeasibility-template.md`, and one\nrecord per experiment under `spikes\u002F`. Report the path, the verdict and what\nstayed open.\n\n## Rules\n\n- **Never invent.** Anything unsettled becomes\n  `[NEEDS CLARIFICATION: what is missing - optionA\u002FoptionB?]`.\n- **Never decide for the developer.** Deliver findings, verdict and a\n  recommendation; the decision is theirs.\n- **Stay inside the appetite.** Research and spikes are spent from it, not\n  added to it.\n- **One page plus the tables.** Depth goes into the spike records.\n- Never edit `specs\u002F`, the constitution or `.specify\u002F`.\n\n## Spec-driven development\n\nThis is Phase 0 research: the verdicts and limits fill the **Technical\nContext** that `\u002Fspeckit.plan` needs (language, dependencies, storage,\nplatform, constraints, scale). Hand `feasibility.md` over and let the command\nwrite `plan.md` and `research.md`.\n",{"name":5,"tags":24,"description":7,"allowed-tools":30},[25,26,27,28,29],"feasibility","research","spike","build-vs-buy","spec-driven-development",[31,32,33,34,35,36,37,38],"Read","Write","Edit","Glob","Grep","Bash","WebSearch","WebFetch","main","0041f4d6875474c146882eae5e8f2d4bc78fca8b","skills\u002Ftech-feasibility",0,"https:\u002F\u002Fgithub.com\u002Fleolicona\u002Fagents-skills","https:\u002F\u002Fgithub.com\u002Fleolicona","Software Engineer | HTML, CSS, TypeScript, React, Next.js | Node.js, Python | AWS ","2026-09-06 10:50:30",{"source":8,"skillName":5,"commands":48,"prompt":58,"zipAvailable":59},[49,52,55],{"tool":50,"command":51},"npx","npx skills add leolicona\u002Fagents-skills --skill tech-feasibility",{"tool":53,"command":54},"bunx","bunx skills add leolicona\u002Fagents-skills --skill tech-feasibility",{"tool":56,"command":57},"pnpm","pnpm dlx skills add leolicona\u002Fagents-skills --skill tech-feasibility","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fleolicona\u002Fagents-skills\u002Fblob\u002Fmain\u002Fskills\u002Ftech-feasibility\u002FSKILL.md",true,[61,67,71],{"path":62,"resourceType":63,"resourceTypeDesc":64,"sizeBytes":65,"sourceHtmlUrl":66},"references\u002Ffeasibility-template.md","REFERENCE","参考文档",1797,"https:\u002F\u002Fgithub.com\u002Fleolicona\u002Fagents-skills\u002Fblob\u002Fmain\u002Fskills\u002Ftech-feasibility\u002Freferences\u002Ffeasibility-template.md",{"path":68,"resourceType":63,"resourceTypeDesc":64,"sizeBytes":69,"sourceHtmlUrl":70},"references\u002Fspike-record.md",1352,"https:\u002F\u002Fgithub.com\u002Fleolicona\u002Fagents-skills\u002Fblob\u002Fmain\u002Fskills\u002Ftech-feasibility\u002Freferences\u002Fspike-record.md",{"path":72,"resourceType":63,"resourceTypeDesc":64,"sizeBytes":73,"sourceHtmlUrl":74},"references\u002Fspike.md",3604,"https:\u002F\u002Fgithub.com\u002Fleolicona\u002Fagents-skills\u002Fblob\u002Fmain\u002Fskills\u002Ftech-feasibility\u002Freferences\u002Fspike.md",[76,83,90],{"skillNo":77,"name":78,"nameZh":6,"description":79,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":80,"categories":81,"keywords":82,"publishedAt":19},"01M7C37RZ15BGFP27PZP51HW3W","demo-con-recursos","Skill de prueba con carpeta references\u002F. Usala para comprobar que la instalacion copia la carpeta completa y no solo SKILL.md.","https:\u002F\u002Fgithub.com\u002Fleolicona\u002Fagents-skills\u002Fblob\u002Fmain\u002Ftest\u002Ffixtures\u002Fcatalog\u002Fdemo-con-recursos\u002FSKILL.md",[],[],{"skillNo":84,"name":85,"nameZh":6,"description":86,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":87,"categories":88,"keywords":89,"publishedAt":19},"01M7C37N9GV67B57YHZ2J21HSP","demo-basica","Skill de prueba minima usada por la suite. Usala solo en las pruebas del CLI, nunca en un proyecto real.","https:\u002F\u002Fgithub.com\u002Fleolicona\u002Fagents-skills\u002Fblob\u002Fmain\u002Ftest\u002Ffixtures\u002Fcatalog\u002Fdemo-basica\u002FSKILL.md",[],[],{"skillNo":91,"name":92,"nameZh":6,"description":93,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":94,"categories":95,"keywords":96,"publishedAt":19},"01M7C37CQNNQ0TSS4M2X686A5B","project-brief","Interviews the user with a short series of questions and turns the answers into a projectbrief.md that feeds GitHub Spec Kit. Use when the user asks to create, start or fill in a project brief or kickoff document, or wants the product context before \u002Fspeckit.constitution and \u002Fspeckit.specify.","https:\u002F\u002Fgithub.com\u002Fleolicona\u002Fagents-skills\u002Fblob\u002Fmain\u002Fskills\u002Fproject-brief\u002FSKILL.md",[],[]]