[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M75NFBEZM8867FE7CZ39MECG":3,"skill-same:01M75NFBEZM8867FE7CZ39MECG":53},{"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":25,"commitSha":26,"skillDirectory":27,"forksCount":14,"repositoryHtmlUrl":28,"repositoryHomepageUrl":29,"ownerHtmlUrl":30,"ownerBlogUrl":6,"ownerDescription":31,"updatedAt":32,"install":33,"resources":46},"01M75NFBEZM8867FE7CZ39MECG","discover",null,"Extract feature intent and requirements through grill-me rounds when Scope cannot state the outcome and acceptance criteria clearly enough. Supports light ideation for a fuzzy concept and structured requirements extraction for complex work. Writes a requirements artifact into the workstream and returns to scope.","don-smith\u002Fmyflow","don-smith","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F228761?v=4","01M75NEQQ3GJBN4ARH0S5EFYV2","Don Smith","01M75NEQQVBT5HAYN363PDG334",0,"MIT","https:\u002F\u002Fgithub.com\u002Fdon-smith\u002Fmyflow\u002Fblob\u002Fmain\u002Fskills\u002Fdiscover\u002FSKILL.md",[],[],"2026-11-12 13:55:43",false,"---\nname: discover\ndescription: Extract feature intent and requirements through grill-me rounds when Scope cannot state the outcome and acceptance criteria clearly enough. Supports light ideation for a fuzzy concept and structured requirements extraction for complex work. Writes a requirements artifact into the workstream and returns to scope.\nargument-hint: \"[free-text feature description | existing artifact path]\"\n---\n\n# Discover\n\n\u003CMyFlow-Note>\nIn MyFlow Stage 1, `scope` is the canonical entry point.\nUse `discover` explicitly when Scope reveals ambiguous intent — the work needs deeper requirements extraction than the adaptive alignment flow provides.\nProduces: `\u003Cworkstream-root>\u002F\u003Cworkstream-id>\u002Fscope\u002F\u003Ctimestamp>_requirements.md`. Returns to: `scope`.\n\u003C\u002FMyFlow-Note>\n\nYou are tasked with extracting feature intent and requirements through `grill-me` rounds, then writing a requirements document that Scope links from its alignment artifact.\n\nDiscover operates on a spectrum: from light ideation for fuzzy concepts through structured requirements extraction for complex work. The output adjusts to the depth needed — a few crisp sections for straightforward work, a full document for high-stakes features.\n\n## Input\n\n`$ARGUMENTS` — free-text feature description, or path to an existing requirements document, ticket, or doc for refinement.\n\n## Metadata\n\nRun this block from this skill's directory and keep its full output. Every path is relative to this installed `skills\u002Fdiscover\u002FSKILL.md`.\n\n```bash\nnode ..\u002F_shared\u002Fnow.mjs\necho\nnode ..\u002F_shared\u002Fgit-context.mjs\n```\n\nCopy values verbatim — do not reformat the timezone offset.\n\n## Flow\n\n1. Input → 2. Intent question → 3. Codebase look → 4. Requirement branches → 5. Grill-me rounds → 6. Synthesize → 7. Write artifact → 8. Return to Scope\n\n## Steps\n\n### Step 1: Input handling\n\n1. **No argument provided**: ask for a free-text feature description, or the path to an existing requirements document, ticket, or doc to refine. Then wait for input.\n\n2. **Detect input shape** — parse the input:\n   - If the argument is an existing file path, read it FULLY. Treat its content as baseline context — the interview surfaces gaps, missing requirements, and unstated assumptions.\n   - Otherwise → fresh-feature mode: the entire argument is the free-text feature description.\n\n3. **Read any other files mentioned** — tickets, docs, related artifacts, explicit `path:line` references — FULLY before proceeding.\n\n**No exploration in Step 1.** Read only the paths the developer named. Grounding starts in Step 3.\n\nEach invocation writes a NEW timestamp-distinct artifact (Step 7). To iterate on a prior one, re-invoke discover with its path.\n\n### Step 2: Foundational intent question\n\nBefore any codebase look, ask the foundational intent question. This is purely conversational — no exploration, no recommendation, no `file:line` citations.\n\n1. **Ask one open-ended intent question**, following the structured-question capability in `..\u002Fmyflow\u002Freferences\u002Fcapabilities.md`:\n   - Frame: \"What problem are you solving and who hits it?\" or \"What does success look like?\" — phrase it for the specific feature.\n   - **No recommended answer.** `grill-me` carries this rule: the developer generates the framing, so an intent question offers open shapes that route the answer, never a recommendation and never a solution shape.\n   - **No `file:line` citations** — the codebase has nothing to say about intent.\n\n2. **Capture the answer in the developer's own words.** This text feeds the Problem and Intent section verbatim.\n\n3. **Probe-readiness check**: does the stated intent support a *narrow* look at the code? If yes → Step 3. If not, ask one more intent question to sharpen scope, then re-check. Cap: 3 intent questions before falling through to Step 3.\n\n### Step 3: Short codebase look\n\nGround the rounds in concrete codebase evidence, shaped by the stated intent.\n\n1. **Keep the slice narrow** — one component, one seam. Locate the relevant code and read what the search surfaces (≤5 files). Use the read-only exploration capability in `..\u002Fmyflow\u002Freferences\u002Fcapabilities.md`.\n\n2. **Cap the look at two passes.** Discover is not research: when a question needs a real investigation, name it for `research` instead.\n\n3. **Empty results are not fatal.** If the look returns little, record \"no codebase precedent\" as evidence.\n\n### Step 4: Requirement branches\n\nBuild the design tree `grill-me` works, rooted in the developer's stated problem. Its immediate branches are the requirement categories this skill owns:\n\n- Goals and non-goals\n- Functional requirements\n- Non-functional requirements — performance, security, UX, accessibility, reliability\n- Constraints and assumptions\n- Acceptance criteria\n- Recommended approach\n\nOrder the branches by dependency: root → goals → constraints → solution shape → details. Mark any branch the Step 3 evidence already settles, with its `file:line` citation, and confirm those pre-resolutions in the first round rather than recording them silently. Expand a branch's children only once it resolves; do not present the tree unless the developer asks for it.\n\n### Step 5: Grill-me rounds\n\nRun the interview as `grill-me` rounds: ask the whole frontier — every branch whose prerequisites are settled — in one numbered round, each question with its recommended answer, then wait for the answers before recomputing the frontier. Intent is already settled in Step 2 and carries no recommendation; every requirement branch is a proposal the developer reviews, so it does carry one. Derive each recommendation from the stated intent, the Step 3 evidence, and project conventions.\n\nClassify each response:\n\n- **Decision**: record it, resolve the branch, continue.\n- **Correction**: look again at the newly named area, then adjust the affected subtree.\n- **Scope adjustment**: update the tree and record the decision. Related but unrequested observations go to Suggested follow-ups.\n- **Cross-cutting answer**: mark the branch cross-cutting and re-queue it under each affected parent.\n- **Defer**: add it to Open questions, resolve by deferral, continue.\n\n**Termination — depth check, not bucket-fill.** Stop when every branch has a decision or a deferral, the developer's own words appear in Problem and Goals, and no decision reads \"recommendation accepted\" without a rationale. Do not invent questions to pad the interview, and never ask a final \"looks good?\" rubber-stamp question.\n\n### Step 6: Synthesize the requirements body\n\nRead `templates\u002Ffrd.md` (relative to this skill folder) at runtime to confirm the section list and frontmatter shape.\n\nCompile the rounds into the document. Redistribute answers into the template's sections:\n\n- **Summary** — 2-3 sentences capturing the settled feature concept.\n- **Problem and intent** — the developer's framing from Step 2, in their own words.\n- **Goals \u002F Non-goals** — explicit in and out lists.\n- **Functional requirements** — numbered, each independently testable.\n- **Non-functional requirements** — performance, security, UX, accessibility, reliability.\n- **Constraints and assumptions** — environmental, technical, schedule, organizational.\n- **Acceptance criteria** — observable pass conditions. Each MUST name a concrete command, output, or visible behavior.\n- **Recommended approach** — 1-2 sentences naming the architectural shape the decisions imply.\n- **Decisions** — the full question and answer log per decision.\n- **Open questions** — only what the developer explicitly deferred.\n- **Suggested follow-ups** — related but out-of-scope items.\n- **References** — input files, mentioned tickets, related artifacts.\n\n### Step 7: Write the artifact\n\n1. **Determine metadata** (from the Metadata block above): `repository:` ← `repo:` label; `branch:` and `commit:` ← matching labels; `date:` and `last_updated:` ← `\u003Ciso>`; author ← `author:`.\n\n2. **Write the document** to:\n\n   ```text\n   \u003Cworkstream-root>\u002F\u003Cworkstream-id>\u002Fscope\u002F\u003Ctimestamp>_requirements.md\n   ```\n\n   Frontmatter `status: ready`. When no workstream exists yet, `scope` creates one first — do not invent a workstream directory here.\n\n### Step 8: Return to Scope\n\nReport the artifact path, the requirement and decision counts, and the open questions. Then hand control back:\n\n```text\nThe `scope` skill with \u003Cworkstream-root>\u002F\u003Cworkstream-id>\u002Fscope\u002F\u003Ctimestamp>_requirements.md\n```\n\n`scope` links the requirements document from the alignment artifact and owns the depth, specialist, and next-action decisions from there. Discover does not select downstream specialists itself.\n\nEach invocation writes a fresh artifact; iterate by re-invoking discover with the prior path. The developer may also edit the document directly.\n\n## Light mode and full mode\n\n- **Light mode** (fuzzy concept, open-ended ideation): fewer rounds, focused on clarifying what the work even is. The output may be compact — problem and intent, goals and non-goals, a few decisions. Enough for `scope` to size the work.\n- **Full mode** (complex work, multiple stakeholders, architectural impact): the full round loop with a complete document. All template sections present.\n\nAsk rounds until the tree resolves. Don't force ceremony for simple work.\n\n## Visual companion\n\nWhen the topic involves visual questions (UI mockups, layout comparisons, architecture diagrams), offer an optional browser-based companion once during the interview:\n\n> \"Some of what we're working on might be easier to explain visually. I can show mockups, diagrams, and comparisons in a browser. Want to try it?\"\n\nThis offer MUST be its own message. If they decline, proceed text-only. If they accept, use the browser for the questions that benefit from visual treatment — not every question needs it.\n\n## Important notes\n\n- **Always interview-first, intent-first**: never write the document without running the rounds.\n- **Intent generates, everything else reviews**: intent is the developer's framing. Every other question is a proposal they review.\n- **Lazy tree, no full-tree pre-build**: build the root and its immediate branches only. Expand each branch's children after it resolves.\n- **Pre-resolutions confirm, never silently record**: put evidence-based branches in the first round.\n- **Interview order is not section order**: walk in dependency order; redistribute in Step 6.\n- **Never write or edit source files**: this skill produces one artifact.\n- **Short look, not research**: the codebase look stays within two passes. Name a real investigation for `research`.\n","# Discover\n\n\u003CMyFlow-Note>\nIn MyFlow Stage 1, `scope` is the canonical entry point.\nUse `discover` explicitly when Scope reveals ambiguous intent — the work needs deeper requirements extraction than the adaptive alignment flow provides.\nProduces: `\u003Cworkstream-root>\u002F\u003Cworkstream-id>\u002Fscope\u002F\u003Ctimestamp>_requirements.md`. Returns to: `scope`.\n\u003C\u002FMyFlow-Note>\n\nYou are tasked with extracting feature intent and requirements through `grill-me` rounds, then writing a requirements document that Scope links from its alignment artifact.\n\nDiscover operates on a spectrum: from light ideation for fuzzy concepts through structured requirements extraction for complex work. The output adjusts to the depth needed — a few crisp sections for straightforward work, a full document for high-stakes features.\n\n## Input\n\n`$ARGUMENTS` — free-text feature description, or path to an existing requirements document, ticket, or doc for refinement.\n\n## Metadata\n\nRun this block from this skill's directory and keep its full output. Every path is relative to this installed `skills\u002Fdiscover\u002FSKILL.md`.\n\n```bash\nnode ..\u002F_shared\u002Fnow.mjs\necho\nnode ..\u002F_shared\u002Fgit-context.mjs\n```\n\nCopy values verbatim — do not reformat the timezone offset.\n\n## Flow\n\n1. Input → 2. Intent question → 3. Codebase look → 4. Requirement branches → 5. Grill-me rounds → 6. Synthesize → 7. Write artifact → 8. Return to Scope\n\n## Steps\n\n### Step 1: Input handling\n\n1. **No argument provided**: ask for a free-text feature description, or the path to an existing requirements document, ticket, or doc to refine. Then wait for input.\n\n2. **Detect input shape** — parse the input:\n   - If the argument is an existing file path, read it FULLY. Treat its content as baseline context — the interview surfaces gaps, missing requirements, and unstated assumptions.\n   - Otherwise → fresh-feature mode: the entire argument is the free-text feature description.\n\n3. **Read any other files mentioned** — tickets, docs, related artifacts, explicit `path:line` references — FULLY before proceeding.\n\n**No exploration in Step 1.** Read only the paths the developer named. Grounding starts in Step 3.\n\nEach invocation writes a NEW timestamp-distinct artifact (Step 7). To iterate on a prior one, re-invoke discover with its path.\n\n### Step 2: Foundational intent question\n\nBefore any codebase look, ask the foundational intent question. This is purely conversational — no exploration, no recommendation, no `file:line` citations.\n\n1. **Ask one open-ended intent question**, following the structured-question capability in `..\u002Fmyflow\u002Freferences\u002Fcapabilities.md`:\n   - Frame: \"What problem are you solving and who hits it?\" or \"What does success look like?\" — phrase it for the specific feature.\n   - **No recommended answer.** `grill-me` carries this rule: the developer generates the framing, so an intent question offers open shapes that route the answer, never a recommendation and never a solution shape.\n   - **No `file:line` citations** — the codebase has nothing to say about intent.\n\n2. **Capture the answer in the developer's own words.** This text feeds the Problem and Intent section verbatim.\n\n3. **Probe-readiness check**: does the stated intent support a *narrow* look at the code? If yes → Step 3. If not, ask one more intent question to sharpen scope, then re-check. Cap: 3 intent questions before falling through to Step 3.\n\n### Step 3: Short codebase look\n\nGround the rounds in concrete codebase evidence, shaped by the stated intent.\n\n1. **Keep the slice narrow** — one component, one seam. Locate the relevant code and read what the search surfaces (≤5 files). Use the read-only exploration capability in `..\u002Fmyflow\u002Freferences\u002Fcapabilities.md`.\n\n2. **Cap the look at two passes.** Discover is not research: when a question needs a real investigation, name it for `research` instead.\n\n3. **Empty results are not fatal.** If the look returns little, record \"no codebase precedent\" as evidence.\n\n### Step 4: Requirement branches\n\nBuild the design tree `grill-me` works, rooted in the developer's stated problem. Its immediate branches are the requirement categories this skill owns:\n\n- Goals and non-goals\n- Functional requirements\n- Non-functional requirements — performance, security, UX, accessibility, reliability\n- Constraints and assumptions\n- Acceptance criteria\n- Recommended approach\n\nOrder the branches by dependency: root → goals → constraints → solution shape → details. Mark any branch the Step 3 evidence already settles, with its `file:line` citation, and confirm those pre-resolutions in the first round rather than recording them silently. Expand a branch's children only once it resolves; do not present the tree unless the developer asks for it.\n\n### Step 5: Grill-me rounds\n\nRun the interview as `grill-me` rounds: ask the whole frontier — every branch whose prerequisites are settled — in one numbered round, each question with its recommended answer, then wait for the answers before recomputing the frontier. Intent is already settled in Step 2 and carries no recommendation; every requirement branch is a proposal the developer reviews, so it does carry one. Derive each recommendation from the stated intent, the Step 3 evidence, and project conventions.\n\nClassify each response:\n\n- **Decision**: record it, resolve the branch, continue.\n- **Correction**: look again at the newly named area, then adjust the affected subtree.\n- **Scope adjustment**: update the tree and record the decision. Related but unrequested observations go to Suggested follow-ups.\n- **Cross-cutting answer**: mark the branch cross-cutting and re-queue it under each affected parent.\n- **Defer**: add it to Open questions, resolve by deferral, continue.\n\n**Termination — depth check, not bucket-fill.** Stop when every branch has a decision or a deferral, the developer's own words appear in Problem and Goals, and no decision reads \"recommendation accepted\" without a rationale. Do not invent questions to pad the interview, and never ask a final \"looks good?\" rubber-stamp question.\n\n### Step 6: Synthesize the requirements body\n\nRead `templates\u002Ffrd.md` (relative to this skill folder) at runtime to confirm the section list and frontmatter shape.\n\nCompile the rounds into the document. Redistribute answers into the template's sections:\n\n- **Summary** — 2-3 sentences capturing the settled feature concept.\n- **Problem and intent** — the developer's framing from Step 2, in their own words.\n- **Goals \u002F Non-goals** — explicit in and out lists.\n- **Functional requirements** — numbered, each independently testable.\n- **Non-functional requirements** — performance, security, UX, accessibility, reliability.\n- **Constraints and assumptions** — environmental, technical, schedule, organizational.\n- **Acceptance criteria** — observable pass conditions. Each MUST name a concrete command, output, or visible behavior.\n- **Recommended approach** — 1-2 sentences naming the architectural shape the decisions imply.\n- **Decisions** — the full question and answer log per decision.\n- **Open questions** — only what the developer explicitly deferred.\n- **Suggested follow-ups** — related but out-of-scope items.\n- **References** — input files, mentioned tickets, related artifacts.\n\n### Step 7: Write the artifact\n\n1. **Determine metadata** (from the Metadata block above): `repository:` ← `repo:` label; `branch:` and `commit:` ← matching labels; `date:` and `last_updated:` ← `\u003Ciso>`; author ← `author:`.\n\n2. **Write the document** to:\n\n   ```text\n   \u003Cworkstream-root>\u002F\u003Cworkstream-id>\u002Fscope\u002F\u003Ctimestamp>_requirements.md\n   ```\n\n   Frontmatter `status: ready`. When no workstream exists yet, `scope` creates one first — do not invent a workstream directory here.\n\n### Step 8: Return to Scope\n\nReport the artifact path, the requirement and decision counts, and the open questions. Then hand control back:\n\n```text\nThe `scope` skill with \u003Cworkstream-root>\u002F\u003Cworkstream-id>\u002Fscope\u002F\u003Ctimestamp>_requirements.md\n```\n\n`scope` links the requirements document from the alignment artifact and owns the depth, specialist, and next-action decisions from there. Discover does not select downstream specialists itself.\n\nEach invocation writes a fresh artifact; iterate by re-invoking discover with the prior path. The developer may also edit the document directly.\n\n## Light mode and full mode\n\n- **Light mode** (fuzzy concept, open-ended ideation): fewer rounds, focused on clarifying what the work even is. The output may be compact — problem and intent, goals and non-goals, a few decisions. Enough for `scope` to size the work.\n- **Full mode** (complex work, multiple stakeholders, architectural impact): the full round loop with a complete document. All template sections present.\n\nAsk rounds until the tree resolves. Don't force ceremony for simple work.\n\n## Visual companion\n\nWhen the topic involves visual questions (UI mockups, layout comparisons, architecture diagrams), offer an optional browser-based companion once during the interview:\n\n> \"Some of what we're working on might be easier to explain visually. I can show mockups, diagrams, and comparisons in a browser. Want to try it?\"\n\nThis offer MUST be its own message. If they decline, proceed text-only. If they accept, use the browser for the questions that benefit from visual treatment — not every question needs it.\n\n## Important notes\n\n- **Always interview-first, intent-first**: never write the document without running the rounds.\n- **Intent generates, everything else reviews**: intent is the developer's framing. Every other question is a proposal they review.\n- **Lazy tree, no full-tree pre-build**: build the root and its immediate branches only. Expand each branch's children after it resolves.\n- **Pre-resolutions confirm, never silently record**: put evidence-based branches in the first round.\n- **Interview order is not section order**: walk in dependency order; redistribute in Step 6.\n- **Never write or edit source files**: this skill produces one artifact.\n- **Short look, not research**: the codebase look stays within two passes. Name a real investigation for `research`.\n",{"name":5,"description":7,"argument-hint":24},"[free-text feature description | existing artifact path]","main","d0a5f620d08689e2cb9684ab07b8055854402f57","skills\u002Fdiscover","https:\u002F\u002Fgithub.com\u002Fdon-smith\u002Fmyflow","https:\u002F\u002Fdon-smith.github.io\u002Fmyflow\u002F","https:\u002F\u002Fgithub.com\u002Fdon-smith","attempting the most positive impact possible while trying to figure out what that means","2026-10-23 09:17:14",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":45},[35,38,41],{"tool":36,"command":37},"npx","npx skills add don-smith\u002Fmyflow --skill discover",{"tool":39,"command":40},"bunx","bunx skills add don-smith\u002Fmyflow --skill discover",{"tool":42,"command":43},"pnpm","pnpm dlx skills add don-smith\u002Fmyflow --skill discover","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fdon-smith\u002Fmyflow\u002Fblob\u002Fmain\u002Fskills\u002Fdiscover\u002FSKILL.md",true,[47],{"path":48,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":51,"sourceHtmlUrl":52},"templates\u002Ffrd.md","REFERENCE","参考文档",3054,"https:\u002F\u002Fgithub.com\u002Fdon-smith\u002Fmyflow\u002Fblob\u002Fmain\u002Fskills\u002Fdiscover\u002Ftemplates\u002Ffrd.md",[54,61,68,75,82,89],{"skillNo":55,"name":56,"nameZh":6,"description":57,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":58,"categories":59,"keywords":60,"publishedAt":19},"01M75NGR4SAVDTSN8A11EW8NBH","clean-support","Use when the clean-stage fixture skill needs a supporting skill to reference.","https:\u002F\u002Fgithub.com\u002Fdon-smith\u002Fmyflow\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fskill-structure\u002Fskills\u002Fclean-support\u002FSKILL.md",[],[],{"skillNo":62,"name":63,"nameZh":6,"description":64,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":65,"categories":66,"keywords":67,"publishedAt":19},"01M75NGPM5WW77DHQN07SSGQWG","clean-stage","Use when exercising the structural skill lint against a tree that satisfies every rule.","https:\u002F\u002Fgithub.com\u002Fdon-smith\u002Fmyflow\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fskill-structure\u002Fskills\u002Fclean-stage\u002FSKILL.md",[],[],{"skillNo":69,"name":70,"nameZh":6,"description":71,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":72,"categories":73,"keywords":74,"publishedAt":19},"01M75NGMXP2ANH3WB2QM5XE75Z","verify","Use after Implement to verify an accepted MyFlow plan, collect review evidence, and produce a workstream-aware Verify report.","https:\u002F\u002Fgithub.com\u002Fdon-smith\u002Fmyflow\u002Fblob\u002Fmain\u002Fskills\u002Fverify\u002FSKILL.md",[],[],{"skillNo":76,"name":77,"nameZh":6,"description":78,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":79,"categories":80,"keywords":81,"publishedAt":19},"01M75NGHZVYBDA346Q14MH7PRG","technical-writing","Layered technical-writing standard: Diátaxis structure, Google developer style sentences, STE instruction rules, Global English syntax. Use when writing or reviewing docs, RFCs, readmes, PR descriptions, or commit messages.","https:\u002F\u002Fgithub.com\u002Fdon-smith\u002Fmyflow\u002Fblob\u002Fmain\u002Fskills\u002Ftechnical-writing\u002FSKILL.md",[],[],{"skillNo":83,"name":84,"nameZh":6,"description":85,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":86,"categories":87,"keywords":88,"publishedAt":19},"01M75NGD25AY0E4F8JRY8AT392","tdd","Use during design and planning to shape testable interfaces, seams, and verification strategy; use during implementation when a plan needs a test-first behavior cycle","https:\u002F\u002Fgithub.com\u002Fdon-smith\u002Fmyflow\u002Fblob\u002Fmain\u002Fskills\u002Ftdd\u002FSKILL.md",[],[],{"skillNo":90,"name":91,"nameZh":6,"description":92,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":93,"categories":94,"keywords":95,"publishedAt":19},"01M75NGB6W0G42QCQNZQSZD0CW","story","Tell a repository's artifact-backed story, answer a question with a dated chapter, or refresh stale story pages.","https:\u002F\u002Fgithub.com\u002Fdon-smith\u002Fmyflow\u002Fblob\u002Fmain\u002Fskills\u002Fstory\u002FSKILL.md",[],[]]