[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M759BTMN1C07PDJBGGC14BGX":3,"skill-same:01M759BTMN1C07PDJBGGC14BGX":67},{"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":26,"commitSha":27,"skillDirectory":28,"forksCount":14,"repositoryHtmlUrl":29,"repositoryHomepageUrl":6,"ownerHtmlUrl":30,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":31,"install":32,"resources":44},"01M759BTMN1C07PDJBGGC14BGX","2-discover",null,"DDD loop 1 — discover: EventStorming timeline, ubiquitous language, hotspots and attendance, tracking confirmed vs candidate and as-is vs to-be vs could-be. Writes docs\u002Fdomain\u002Fdiscovery\u002F.","baodq97\u002Fgovkit","baodq97","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F50482002?v=4","01M759BJ9XC68CW9PQCFK22H1X","Bao Do","01M759BJAY1W696G8A8N841875",0,"MIT","https:\u002F\u002Fgithub.com\u002Fbaodq97\u002Fgovkit\u002Fblob\u002Fmain\u002Fdocs\u002Fresearch\u002Fddd-skill-evals\u002Fskill-snapshot-pre-context-pack\u002F2-discover\u002FSKILL.md",[],[],"2026-11-12 13:55:43",false,"---\nname: 2-discover\ndescription: >\n  DDD loop 1 — discover: EventStorming timeline, ubiquitous language, hotspots and attendance, tracking confirmed vs candidate and as-is vs to-be vs could-be. Writes docs\u002Fdomain\u002Fdiscovery\u002F.\ndisable-model-invocation: true\n---\n\n# Domain Discover\n\n> *\"This is the most crucial aspect of DDD. You cannot skip discovery. If your whole team doesn't\n> build up a good understanding of the domain, all software decisions will be misguided.\"*\n> — ddd-crew, Discover\n\nDiscovery is where a team stops guessing. The output is not a document — it is **shared\nunderstanding**, made durable enough that the next step can build on it.\n\n**What this skill is, honestly:** a facilitator's harness. It prepares the surface, proposes\ncandidates from whatever is written down, asks the questions that surface behaviour, and records\nwhat people say with attribution. It **cannot** replace the people in the room. A run with no\ndomain expert present has discovered nothing; it has only restated the documents.\n\n## Mode detection — do this first\n\n| Signal | Mode |\n|---|---|\n| No repo, or no domain artifacts | **INTERVIEW** — elicit the timeline from scratch |\n| Repo has PRDs, ADRs, schemas, specs, a domain layer | **DISCOVER** — mine candidates first, then interview only the gaps |\n| `docs\u002Fdomain\u002Fdiscovery\u002F` exists | **UPDATE** — discovery is continuous; merge a delta |\n\nThe distinction matters because the two modes fail differently. INTERVIEW risks an unbounded\nconversation, so it needs the stop rule. DISCOVER risks the opposite — mistaking *documented* for\n*discovered*, and shipping a timeline that merely paraphrases a schema. Guard against that by\ntracking, per event, whether a human confirmed it or an artifact implied it.\n\n## Reference files (read as needed)\n\n- `references\u002Feventstorming.md` — the sticky grammar (event, command, actor, policy, read model,\n  external system, hotspot), the three levels (big picture → process → design), and how to run\n  each. Read before facilitating.\n- `references\u002Fother-techniques.md` — Domain Storytelling, Example Mapping, User Journey Mapping,\n  and which question each answers better than EventStorming.\n- `references\u002Finterview-guide.md` — the question sets that surface events, rules and vocabulary,\n  and the stop rule.\n- `references\u002Foutput-template.md` — the exact output contract.\n\n## Who to involve\n\nSay this out loud before starting. The skill's value is capped by who is present:\n\n- people who design, build and test software\n- people who have domain knowledge\n- people who understand product and business strategy\n- people who understand the customers' needs and problems\n- **real end users**\n\nRecord who actually attended. When a group is missing, the areas that depended on them stay\nhotspots rather than becoming confident findings.\n\n## The visual surface\n\nDiscovery is visual and collaborative or it is not discovery. Start the shared surface first:\n\n```bash\nnode \u003Cplugin>\u002Fskills\u002Fview\u002Fscripts\u002Fpreview-server.cjs --dir .ddd-flow\u002Fdiscovery\n```\n\nWrite the model to `\u003Cdir>\u002Fmodel.json`; the page updates itself, so participants watch the timeline\nform instead of waiting for a document. Their clicks land in `events.jsonl`, which you read on your\nnext turn and merge with what they say in conversation.\n\nIf a browser is genuinely unavailable, fall back to a markdown timeline and say that a text\ntimeline loses the thing that makes EventStorming work — everyone seeing the same wall at once.\n\n## Process\n\n### 1. Ground (DISCOVER mode)\n\nRead what exists — PRDs, specs, ADRs, schemas, domain layers, prior `docs\u002Fdomain\u002F`. Extract\n**candidate** events, commands, actors and terms, each tagged with its source. These are proposals\nto be confirmed or corrected by people, never findings. Mark every one `status: candidate`.\n\nSkip to step 2 in INTERVIEW mode.\n\n### 2. Big-picture EventStorming — the timeline\n\nLay out **domain events** in past tense, in rough time order. Start with the event that matters\nmost commercially and work outward; a chronological start from \"user signs up\" wastes the room's\nbest energy on the least interesting part.\n\nFor each event, elicit the surrounding grammar (see the reference): what **command** caused it,\nwhich **actor** issued it, which **policy** reacts to it, what **read model** someone looks at to\ndecide, which **external system** is involved.\n\n**Hotspots** — disagreement, \"it depends\", \"ask X\", visible discomfort — are the most valuable\noutput of the session. Capture them as hotspots; do not resolve them in the room by picking a side.\n\n### 3. Interview the gaps\n\nAsk only where the timeline has a hole. The question sets in `references\u002Finterview-guide.md`\nfollow four rules that matter more than the questions themselves:\n\n- **Ground first, ask second.** Never ask what the documents already answer.\n- **One question at a time.** A wall of questions gets a wall of shallow answers.\n- **Concrete scenarios over abstractions.** *\"Tell me about the last time a transfer was\n  rejected\"* surfaces rules that *\"how does the transfer process work\"* never will. This is the\n  DDD doctrine \"focus on concrete scenarios\" applied to elicitation.\n- **Ask what goes wrong.** Business rules hide behind incident stories. *\"What would happen if two\n  people booked the same unit?\"* is how an invariant gets stated out loud for the first time.\n\n### 4. Ubiquitous language\n\nRecord every term people actually use, with the definition **they** gave. When the same word means\ntwo things to two people, that is not a naming problem to tidy up — it is the strongest boundary\nsignal discovery produces. Record both meanings and who holds each.\n\n### 5. Stop\n\nStop when the next question would not change the model. Discovery is continuous, not infinite: the\nsession ends, the practice does not. Record what is still unknown so the next round starts there\nrather than repeating this one.\n\n### 6. Emit\n\nWrite `docs\u002Fdomain\u002Fdiscovery\u002F` per `references\u002Foutput-template.md`: the event timeline, the\nubiquitous language, the hotspots, and the attendance record. `status: draft`, `owner: TBD`.\n\nThen say plainly what the next step is: `3-decompose` consumes this as its step-2 input, and\nit will be as good as this discovery was.\n\n## Hard rules\n\n- **Give every hotspot a stable id `H1`, `H2`, … and never renumber them.** Hotspots are the one\n  thing every downstream artifact points back at; a boundary justified by \"blocked on H1\" is\n  only checkable while H1 keeps meaning the same question. Re-running this step adds ids, it\n  does not renumber the existing ones.\n- **Length budget: ≤ 120 lines per file in `discovery\u002F`.** A budget caps prose, not findings:\n  over it, cut rationale a reader can infer and anything restated from an upstream artifact —\n  never open questions, provenance, or a stated absence.\n- **Never invent an event, rule, or actor.** Naming `TransferRequested` from \"people request\n  transfers\" is the job. Inventing `TransferCancelled` when nothing in the domain ever cancels is\n  fabrication, and it is undetectable downstream. If the timeline has a gap, the gap is a hotspot.\n- **Distinguish confirmed from candidate.** Every item carries whether a human confirmed it or an\n  artifact implied it. Without this, a run that merely re-read the schemas looks identical to one\n  that talked to the business — and only one of those is discovery.\n- **Distinguish as-is from to-be from could-be**, in a column of its own. What the business does\n  today, what it has decided to change, and what someone floated in the room are three different\n  claims, and a wall records them in the same handwriting. This is a second axis, not a rename of\n  the first: an element can be `confirmed` (a person said it) and `could-be` (what they confirmed is\n  that it is only an idea). Default `as-is`, change it on evidence, and make an unplaceable element\n  a hotspot rather than a guess.\n- **Attribute.** Who said it, when. Terms especially: a definition without a holder cannot be\n  challenged later.\n- **Don't draw boundaries.** Clustering events into candidate contexts is `3-decompose`'s\n  job. Doing it here collapses discovery and design into one step and loses the disagreement that\n  discovery exists to surface.\n- **Don't resolve hotspots to keep things tidy.** An open hotspot is a finding. A quietly closed\n  one is a decision nobody made.\n\n## Worked example\n\n**Input (DISCOVER mode):** an equipment-rental repo whose only domain writing is a set of schema\nmigration ADRs.\n\n**Ground pass** produced candidates from table names: `ReservationCreated`, `InvoiceIssued` —\nboth marked `candidate`, sourced to ADR-0012.\n\n**Interview, one question at a time:**\n\n> *\"Tell me about the last time a booking went wrong.\"*\n> — \"A unit got double-booked across two depots. It's the thing that hurts most.\"\n\nThat answer produced, in one exchange: a **confirmed** event (`EquipmentAllocated`), an\n**invariant** stated for the first time (*\"the same physical unit can never be committed twice for\noverlapping windows, even from a different depot\"*), and a **hotspot** (nobody agreed who releases\na unit when the depot changes mid-rental).\n\n**Ubiquitous-language conflict found:** *\"Transfer\"* meant a physical depot-to-depot move to\noperations, and a billing line item to finance. Both recorded, with holders. That single conflict\nis worth more to the next step than the whole candidate list from the ground pass.\n\nNote what the example does **not** do: it doesn't promote the two candidate events to confirmed\njust because they came from an accepted ADR, and it doesn't settle who releases the unit — that\nstays a hotspot with a name attached.\n","# Domain Discover\n\n> *\"This is the most crucial aspect of DDD. You cannot skip discovery. If your whole team doesn't\n> build up a good understanding of the domain, all software decisions will be misguided.\"*\n> — ddd-crew, Discover\n\nDiscovery is where a team stops guessing. The output is not a document — it is **shared\nunderstanding**, made durable enough that the next step can build on it.\n\n**What this skill is, honestly:** a facilitator's harness. It prepares the surface, proposes\ncandidates from whatever is written down, asks the questions that surface behaviour, and records\nwhat people say with attribution. It **cannot** replace the people in the room. A run with no\ndomain expert present has discovered nothing; it has only restated the documents.\n\n## Mode detection — do this first\n\n| Signal | Mode |\n|---|---|\n| No repo, or no domain artifacts | **INTERVIEW** — elicit the timeline from scratch |\n| Repo has PRDs, ADRs, schemas, specs, a domain layer | **DISCOVER** — mine candidates first, then interview only the gaps |\n| `docs\u002Fdomain\u002Fdiscovery\u002F` exists | **UPDATE** — discovery is continuous; merge a delta |\n\nThe distinction matters because the two modes fail differently. INTERVIEW risks an unbounded\nconversation, so it needs the stop rule. DISCOVER risks the opposite — mistaking *documented* for\n*discovered*, and shipping a timeline that merely paraphrases a schema. Guard against that by\ntracking, per event, whether a human confirmed it or an artifact implied it.\n\n## Reference files (read as needed)\n\n- `references\u002Feventstorming.md` — the sticky grammar (event, command, actor, policy, read model,\n  external system, hotspot), the three levels (big picture → process → design), and how to run\n  each. Read before facilitating.\n- `references\u002Fother-techniques.md` — Domain Storytelling, Example Mapping, User Journey Mapping,\n  and which question each answers better than EventStorming.\n- `references\u002Finterview-guide.md` — the question sets that surface events, rules and vocabulary,\n  and the stop rule.\n- `references\u002Foutput-template.md` — the exact output contract.\n\n## Who to involve\n\nSay this out loud before starting. The skill's value is capped by who is present:\n\n- people who design, build and test software\n- people who have domain knowledge\n- people who understand product and business strategy\n- people who understand the customers' needs and problems\n- **real end users**\n\nRecord who actually attended. When a group is missing, the areas that depended on them stay\nhotspots rather than becoming confident findings.\n\n## The visual surface\n\nDiscovery is visual and collaborative or it is not discovery. Start the shared surface first:\n\n```bash\nnode \u003Cplugin>\u002Fskills\u002Fview\u002Fscripts\u002Fpreview-server.cjs --dir .ddd-flow\u002Fdiscovery\n```\n\nWrite the model to `\u003Cdir>\u002Fmodel.json`; the page updates itself, so participants watch the timeline\nform instead of waiting for a document. Their clicks land in `events.jsonl`, which you read on your\nnext turn and merge with what they say in conversation.\n\nIf a browser is genuinely unavailable, fall back to a markdown timeline and say that a text\ntimeline loses the thing that makes EventStorming work — everyone seeing the same wall at once.\n\n## Process\n\n### 1. Ground (DISCOVER mode)\n\nRead what exists — PRDs, specs, ADRs, schemas, domain layers, prior `docs\u002Fdomain\u002F`. Extract\n**candidate** events, commands, actors and terms, each tagged with its source. These are proposals\nto be confirmed or corrected by people, never findings. Mark every one `status: candidate`.\n\nSkip to step 2 in INTERVIEW mode.\n\n### 2. Big-picture EventStorming — the timeline\n\nLay out **domain events** in past tense, in rough time order. Start with the event that matters\nmost commercially and work outward; a chronological start from \"user signs up\" wastes the room's\nbest energy on the least interesting part.\n\nFor each event, elicit the surrounding grammar (see the reference): what **command** caused it,\nwhich **actor** issued it, which **policy** reacts to it, what **read model** someone looks at to\ndecide, which **external system** is involved.\n\n**Hotspots** — disagreement, \"it depends\", \"ask X\", visible discomfort — are the most valuable\noutput of the session. Capture them as hotspots; do not resolve them in the room by picking a side.\n\n### 3. Interview the gaps\n\nAsk only where the timeline has a hole. The question sets in `references\u002Finterview-guide.md`\nfollow four rules that matter more than the questions themselves:\n\n- **Ground first, ask second.** Never ask what the documents already answer.\n- **One question at a time.** A wall of questions gets a wall of shallow answers.\n- **Concrete scenarios over abstractions.** *\"Tell me about the last time a transfer was\n  rejected\"* surfaces rules that *\"how does the transfer process work\"* never will. This is the\n  DDD doctrine \"focus on concrete scenarios\" applied to elicitation.\n- **Ask what goes wrong.** Business rules hide behind incident stories. *\"What would happen if two\n  people booked the same unit?\"* is how an invariant gets stated out loud for the first time.\n\n### 4. Ubiquitous language\n\nRecord every term people actually use, with the definition **they** gave. When the same word means\ntwo things to two people, that is not a naming problem to tidy up — it is the strongest boundary\nsignal discovery produces. Record both meanings and who holds each.\n\n### 5. Stop\n\nStop when the next question would not change the model. Discovery is continuous, not infinite: the\nsession ends, the practice does not. Record what is still unknown so the next round starts there\nrather than repeating this one.\n\n### 6. Emit\n\nWrite `docs\u002Fdomain\u002Fdiscovery\u002F` per `references\u002Foutput-template.md`: the event timeline, the\nubiquitous language, the hotspots, and the attendance record. `status: draft`, `owner: TBD`.\n\nThen say plainly what the next step is: `3-decompose` consumes this as its step-2 input, and\nit will be as good as this discovery was.\n\n## Hard rules\n\n- **Give every hotspot a stable id `H1`, `H2`, … and never renumber them.** Hotspots are the one\n  thing every downstream artifact points back at; a boundary justified by \"blocked on H1\" is\n  only checkable while H1 keeps meaning the same question. Re-running this step adds ids, it\n  does not renumber the existing ones.\n- **Length budget: ≤ 120 lines per file in `discovery\u002F`.** A budget caps prose, not findings:\n  over it, cut rationale a reader can infer and anything restated from an upstream artifact —\n  never open questions, provenance, or a stated absence.\n- **Never invent an event, rule, or actor.** Naming `TransferRequested` from \"people request\n  transfers\" is the job. Inventing `TransferCancelled` when nothing in the domain ever cancels is\n  fabrication, and it is undetectable downstream. If the timeline has a gap, the gap is a hotspot.\n- **Distinguish confirmed from candidate.** Every item carries whether a human confirmed it or an\n  artifact implied it. Without this, a run that merely re-read the schemas looks identical to one\n  that talked to the business — and only one of those is discovery.\n- **Distinguish as-is from to-be from could-be**, in a column of its own. What the business does\n  today, what it has decided to change, and what someone floated in the room are three different\n  claims, and a wall records them in the same handwriting. This is a second axis, not a rename of\n  the first: an element can be `confirmed` (a person said it) and `could-be` (what they confirmed is\n  that it is only an idea). Default `as-is`, change it on evidence, and make an unplaceable element\n  a hotspot rather than a guess.\n- **Attribute.** Who said it, when. Terms especially: a definition without a holder cannot be\n  challenged later.\n- **Don't draw boundaries.** Clustering events into candidate contexts is `3-decompose`'s\n  job. Doing it here collapses discovery and design into one step and loses the disagreement that\n  discovery exists to surface.\n- **Don't resolve hotspots to keep things tidy.** An open hotspot is a finding. A quietly closed\n  one is a decision nobody made.\n\n## Worked example\n\n**Input (DISCOVER mode):** an equipment-rental repo whose only domain writing is a set of schema\nmigration ADRs.\n\n**Ground pass** produced candidates from table names: `ReservationCreated`, `InvoiceIssued` —\nboth marked `candidate`, sourced to ADR-0012.\n\n**Interview, one question at a time:**\n\n> *\"Tell me about the last time a booking went wrong.\"*\n> — \"A unit got double-booked across two depots. It's the thing that hurts most.\"\n\nThat answer produced, in one exchange: a **confirmed** event (`EquipmentAllocated`), an\n**invariant** stated for the first time (*\"the same physical unit can never be committed twice for\noverlapping windows, even from a different depot\"*), and a **hotspot** (nobody agreed who releases\na unit when the depot changes mid-rental).\n\n**Ubiquitous-language conflict found:** *\"Transfer\"* meant a physical depot-to-depot move to\noperations, and a billing line item to finance. Both recorded, with holders. That single conflict\nis worth more to the next step than the whole candidate list from the ground pass.\n\nNote what the example does **not** do: it doesn't promote the two candidate events to confirmed\njust because they came from an accepted ADR, and it doesn't settle who releases the unit — that\nstays a hotspot with a name attached.\n",{"name":5,"description":24,"disable-model-invocation":25},"DDD loop 1 — discover: EventStorming timeline, ubiquitous language, hotspots and attendance, tracking confirmed vs candidate and as-is vs to-be vs could-be. Writes docs\u002Fdomain\u002Fdiscovery\u002F.\n",true,"main","850634e254ea6a5034bf2eddad5fe95091713231","docs\u002Fresearch\u002Fddd-skill-evals\u002Fskill-snapshot-pre-context-pack\u002F2-discover","https:\u002F\u002Fgithub.com\u002Fbaodq97\u002Fgovkit","https:\u002F\u002Fgithub.com\u002Fbaodq97","2026-09-07 14:26:00",{"source":8,"skillName":5,"commands":33,"prompt":43,"zipAvailable":25},[34,37,40],{"tool":35,"command":36},"npx","npx skills add baodq97\u002Fgovkit --skill 2-discover",{"tool":38,"command":39},"bunx","bunx skills add baodq97\u002Fgovkit --skill 2-discover",{"tool":41,"command":42},"pnpm","pnpm dlx skills add baodq97\u002Fgovkit --skill 2-discover","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fbaodq97\u002Fgovkit\u002Fblob\u002Fmain\u002Fdocs\u002Fresearch\u002Fddd-skill-evals\u002Fskill-snapshot-pre-context-pack\u002F2-discover\u002FSKILL.md",[45,51,55,59,63],{"path":46,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":49,"sourceHtmlUrl":50},"evals\u002Fevals.json","REFERENCE","参考文档",3384,"https:\u002F\u002Fgithub.com\u002Fbaodq97\u002Fgovkit\u002Fblob\u002Fmain\u002Fdocs\u002Fresearch\u002Fddd-skill-evals\u002Fskill-snapshot-pre-context-pack\u002F2-discover\u002Fevals\u002Fevals.json",{"path":52,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":53,"sourceHtmlUrl":54},"references\u002Feventstorming.md",5671,"https:\u002F\u002Fgithub.com\u002Fbaodq97\u002Fgovkit\u002Fblob\u002Fmain\u002Fdocs\u002Fresearch\u002Fddd-skill-evals\u002Fskill-snapshot-pre-context-pack\u002F2-discover\u002Freferences\u002Feventstorming.md",{"path":56,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":57,"sourceHtmlUrl":58},"references\u002Finterview-guide.md",4537,"https:\u002F\u002Fgithub.com\u002Fbaodq97\u002Fgovkit\u002Fblob\u002Fmain\u002Fdocs\u002Fresearch\u002Fddd-skill-evals\u002Fskill-snapshot-pre-context-pack\u002F2-discover\u002Freferences\u002Finterview-guide.md",{"path":60,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":61,"sourceHtmlUrl":62},"references\u002Fother-techniques.md",4402,"https:\u002F\u002Fgithub.com\u002Fbaodq97\u002Fgovkit\u002Fblob\u002Fmain\u002Fdocs\u002Fresearch\u002Fddd-skill-evals\u002Fskill-snapshot-pre-context-pack\u002F2-discover\u002Freferences\u002Fother-techniques.md",{"path":64,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":65,"sourceHtmlUrl":66},"references\u002Foutput-template.md",7229,"https:\u002F\u002Fgithub.com\u002Fbaodq97\u002Fgovkit\u002Fblob\u002Fmain\u002Fdocs\u002Fresearch\u002Fddd-skill-evals\u002Fskill-snapshot-pre-context-pack\u002F2-discover\u002Freferences\u002Foutput-template.md",[68,75,82,89,96,102],{"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},"01M759HTCHQJVZ4JWDA2JNPJJZ","trigger-shaped","Use when the user says \"plan a sprint retro\"; drafts retro\u002Fnotes.md for the team.","https:\u002F\u002Fgithub.com\u002Fbaodq97\u002Fgovkit\u002Fblob\u002Fmain\u002Fscripts\u002Ffixtures\u002Fskill-lint\u002Ftriggering\u002Fskills\u002Ftrigger-shaped\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},"01M759HS3E2T97ZN1ED2Z96WAJ","terse-unguarded","Step 2 — writes docs\u002Fx\u002Foutput.json quietly.","https:\u002F\u002Fgithub.com\u002Fbaodq97\u002Fgovkit\u002Fblob\u002Fmain\u002Fscripts\u002Ffixtures\u002Fskill-lint\u002Ftriggering\u002Fskills\u002Fterse-unguarded\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},"01M759HPZN28VZHKYFR754CVXT","guarded","Summarizes ledger totals into reports\u002Fmonthly.csv.","https:\u002F\u002Fgithub.com\u002Fbaodq97\u002Fgovkit\u002Fblob\u002Fmain\u002Fscripts\u002Ffixtures\u002Fskill-lint\u002Ftriggering\u002Fskills\u002Fguarded\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},"01M759HMFQAZAWFVSS3KWPCCW0","open","Draft a spec. Use when asked to \"draft a spec\" or \"write the RFC\".","https:\u002F\u002Fgithub.com\u002Fbaodq97\u002Fgovkit\u002Fblob\u002Fmain\u002Fscripts\u002Ffixtures\u002Fskill-lint\u002Fdead-hint\u002Fskills\u002Fopen\u002FSKILL.md",[],[],{"skillNo":97,"name":84,"nameZh":6,"description":98,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":99,"categories":100,"keywords":101,"publishedAt":19},"01M759HJ3XX74S70PW9AR8NS66","Deploy the thing.","https:\u002F\u002Fgithub.com\u002Fbaodq97\u002Fgovkit\u002Fblob\u002Fmain\u002Fscripts\u002Ffixtures\u002Fskill-lint\u002Fdead-hint\u002Fskills\u002Fguarded\u002FSKILL.md",[],[],{"skillNo":103,"name":104,"nameZh":6,"description":105,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":106,"categories":107,"keywords":108,"publishedAt":19},"01M759HCWJN9ACRNC1JW76DR1E","working-discipline","Thinking-checkpoint discipline for substantive implementation, debugging, or autonomous work — 21 trigger-gated items (trigger → mandatory question → evidence to produce) that cut agent error and human escalation. Use before an autonomous or long-running run, when asked to \"work carefully\", \"double-check\", \"làm cẩn thận\", \"reduce errors\", \"be more autonomous\", or after looping twice on the same failing fix. NOT a procedure — checkpoints fire on their triggers and demand evidence, with an N\u002FA escape valve so a strong model pays near-zero overhead.","https:\u002F\u002Fgithub.com\u002Fbaodq97\u002Fgovkit\u002Fblob\u002Fmain\u002Fplugins\u002Fswe-flow\u002Fskills\u002Fworking-discipline\u002FSKILL.md",[],[]]