[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7B4AD0ZJYZ3HAJCCBM4CPTY":3,"skill-same:01M7B4AD0ZJYZ3HAJCCBM4CPTY":52},{"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":27,"repositoryHtmlUrl":28,"repositoryHomepageUrl":6,"ownerHtmlUrl":29,"ownerBlogUrl":6,"ownerDescription":30,"updatedAt":31,"install":32,"resources":45},"01M7B4AD0ZJYZ3HAJCCBM4CPTY","write-a-spec",null,"Turn an accepted intent into a spec an engineer can build from in this repository: the goal restated, the approach and why, every path it touches, the steps, the tests that prove it, the risks, and what is still open. For a step that designs a change before anyone builds it.","emyann\u002Fnemia","emyann","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F4748419?v=4","01M7B458J23KJ5S5AEBR86G7RN","Yann RENAUDIN","01M7B458JPM0CCBDP4KCR6NATT",3,"MIT","https:\u002F\u002Fgithub.com\u002Femyann\u002Fnemia\u002Fblob\u002Fmain\u002Fpackages\u002Ftemplates\u002Fsdlc\u002Fskills\u002Fwrite-a-spec\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: write-a-spec\ndescription: \"Turn an accepted intent into a spec an engineer can build from in this repository: the goal restated, the approach and why, every path it touches, the steps, the tests that prove it, the risks, and what is still open. For a step that designs a change before anyone builds it.\"\n---\n\n# Write a spec\n\nThe intent says what should be true; the spec says how this repository gets there. The engineer and the reviewer read the spec, not the\nintent, so restate the intent's goal and constraints in it.\n\n1. **Read the intent**, under *From run* in your brief, then the code it touches. Facts are yours to find: read, search, run the suite, or\n   dispatch a sub-agent for a wide search. Never ask the person for what you can look up.\n2. **Read the rules the change meets**: the repository's instruction files (`CLAUDE.md` at the root and in each folder the change\n   touches) and every architecture or design document they name — its boundaries and what each owns, its size limits (a limit is met\n   by cutting, never by raising a ceiling), how to test.\n3. **Choose the approach**, and say why this way rather than the obvious other. Challenge any mechanism the intent asserts: read the\n   code that would carry it.\n4. **Place each change** by the boundaries the repository's own rules define, in their terms. Each change says `where` it lives and\n   `why` there and not in the neighbouring place, citing the rule. The `placement` table gives each concept the change adds or moves:\n   where it lives, the rule that puts it there, and the alternative rejected. A placement the rules do not settle is the person's:\n   ask it in a round.\n5. **List every change by path**, the **steps in order**, and the **tests**: for each rule the change adds, the fixture that makes it\n   refuse, and the check that it fails without the rule. **Name the one layer that proves each rule, and ask for it there alone**: a\n   proof where the repository proves its pure decisions, a schema where the rule is a boundary's shape, else the cheapest test that\n   can fail for it. A test of the running whole, in a browser or end to end, is for what only the running whole shows: an\n   interaction, a layout, two parts wired together. A rule restated at every level is three tests to change for one fact, and every\n   round after pays for each.\n6. **Cut it by its seams, never by its cost.** Cut the work into parts by where each lives: a layer, a package or a door, the\n   boundaries your placement already names. Each part is a distinct area, says what it needs of the others, and is ordered only where\n   one needs another, so parts with no need of each other can be built at the same time. Propose a second intent only for a seam of\n   logic, where a part cannot be built until another has landed and been used; never cut to stay under a budget. When the whole costs\n   more than a run usually does, say so, and size each step's budget to it.\n7. **Ask what is the person's.** A choice the intent leaves open, a trade-off only they can weigh: ask it in a round, numbered, each with\n   your recommended answer. Record each answer, and who gave it, in `decisions`.\n\nRead a file with Read, list a folder with Glob and search with Grep; never `grep` or a pipe in\nBash. `cat`, `head`, `tail`, `wc`, `cut` and `ls` read a file or a folder inside the workspace, and your fence refuses each one a path outside it; never `sed`, `awk`, `find`, `sort`, `uniq` or `tee`, which can write a file or run a program, nor `wc --files0-from`, which reads the paths a file lists. Otherwise Bash runs the checks below and git's read-only questions — `git diff`, `git log`, `git show`, `git status`, `git blame`,\n`git grep \u003Cpattern> \u003Crev>`, `git ls-files`, `git rev-parse`, `git merge-base` and the rest of the read-only set your fence allows — each\nbare and one per call, and never a write. Read, Grep and Glob reach this workspace, by paths from its root, and each folder your brief says you\nmay read, under its own path; never another run's folder, the store (`~\u002F.agent-workflows`), `~\u002F.claude`, `~\u002F.cache` or a system path:\nyour fence refuses them without asking. What your brief quotes from outside them is all you have of it: ask in a round for anything more.\n\nRun each check your step names bare, one command per call: never wrapped in `timeout` or `( … )`, piped to `tail` or `grep`, or\nredirected; paths from the workspace root, never `cd`, `git -C`, `VAR=` or `for`. Anything else is refused.\n\nA spec that reads well and names a function that does not exist is worse than none: name only what you read.\n","# Write a spec\n\nThe intent says what should be true; the spec says how this repository gets there. The engineer and the reviewer read the spec, not the\nintent, so restate the intent's goal and constraints in it.\n\n1. **Read the intent**, under *From run* in your brief, then the code it touches. Facts are yours to find: read, search, run the suite, or\n   dispatch a sub-agent for a wide search. Never ask the person for what you can look up.\n2. **Read the rules the change meets**: the repository's instruction files (`CLAUDE.md` at the root and in each folder the change\n   touches) and every architecture or design document they name — its boundaries and what each owns, its size limits (a limit is met\n   by cutting, never by raising a ceiling), how to test.\n3. **Choose the approach**, and say why this way rather than the obvious other. Challenge any mechanism the intent asserts: read the\n   code that would carry it.\n4. **Place each change** by the boundaries the repository's own rules define, in their terms. Each change says `where` it lives and\n   `why` there and not in the neighbouring place, citing the rule. The `placement` table gives each concept the change adds or moves:\n   where it lives, the rule that puts it there, and the alternative rejected. A placement the rules do not settle is the person's:\n   ask it in a round.\n5. **List every change by path**, the **steps in order**, and the **tests**: for each rule the change adds, the fixture that makes it\n   refuse, and the check that it fails without the rule. **Name the one layer that proves each rule, and ask for it there alone**: a\n   proof where the repository proves its pure decisions, a schema where the rule is a boundary's shape, else the cheapest test that\n   can fail for it. A test of the running whole, in a browser or end to end, is for what only the running whole shows: an\n   interaction, a layout, two parts wired together. A rule restated at every level is three tests to change for one fact, and every\n   round after pays for each.\n6. **Cut it by its seams, never by its cost.** Cut the work into parts by where each lives: a layer, a package or a door, the\n   boundaries your placement already names. Each part is a distinct area, says what it needs of the others, and is ordered only where\n   one needs another, so parts with no need of each other can be built at the same time. Propose a second intent only for a seam of\n   logic, where a part cannot be built until another has landed and been used; never cut to stay under a budget. When the whole costs\n   more than a run usually does, say so, and size each step's budget to it.\n7. **Ask what is the person's.** A choice the intent leaves open, a trade-off only they can weigh: ask it in a round, numbered, each with\n   your recommended answer. Record each answer, and who gave it, in `decisions`.\n\nRead a file with Read, list a folder with Glob and search with Grep; never `grep` or a pipe in\nBash. `cat`, `head`, `tail`, `wc`, `cut` and `ls` read a file or a folder inside the workspace, and your fence refuses each one a path outside it; never `sed`, `awk`, `find`, `sort`, `uniq` or `tee`, which can write a file or run a program, nor `wc --files0-from`, which reads the paths a file lists. Otherwise Bash runs the checks below and git's read-only questions — `git diff`, `git log`, `git show`, `git status`, `git blame`,\n`git grep \u003Cpattern> \u003Crev>`, `git ls-files`, `git rev-parse`, `git merge-base` and the rest of the read-only set your fence allows — each\nbare and one per call, and never a write. Read, Grep and Glob reach this workspace, by paths from its root, and each folder your brief says you\nmay read, under its own path; never another run's folder, the store (`~\u002F.agent-workflows`), `~\u002F.claude`, `~\u002F.cache` or a system path:\nyour fence refuses them without asking. What your brief quotes from outside them is all you have of it: ask in a round for anything more.\n\nRun each check your step names bare, one command per call: never wrapped in `timeout` or `( … )`, piped to `tail` or `grep`, or\nredirected; paths from the workspace root, never `cd`, `git -C`, `VAR=` or `for`. Anything else is refused.\n\nA spec that reads well and names a function that does not exist is worse than none: name only what you read.\n",{"name":5,"description":7},"main","e3efae8a132cdd1bb19337ef25a64fc7444f2708","packages\u002Ftemplates\u002Fsdlc\u002Fskills\u002Fwrite-a-spec",0,"https:\u002F\u002Fgithub.com\u002Femyann\u002Fnemia","https:\u002F\u002Fgithub.com\u002Femyann","Senior Software Engineer","2026-11-15 05:55:45",{"source":8,"skillName":5,"commands":33,"prompt":43,"zipAvailable":44},[34,37,40],{"tool":35,"command":36},"npx","npx skills add emyann\u002Fnemia --skill write-a-spec",{"tool":38,"command":39},"bunx","bunx skills add emyann\u002Fnemia --skill write-a-spec",{"tool":41,"command":42},"pnpm","pnpm dlx skills add emyann\u002Fnemia --skill write-a-spec","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Femyann\u002Fnemia\u002Fblob\u002Fmain\u002Fpackages\u002Ftemplates\u002Fsdlc\u002Fskills\u002Fwrite-a-spec\u002FSKILL.md",true,[46],{"path":47,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":50,"sourceHtmlUrl":51},"rules.json","REFERENCE","参考文档",945,"https:\u002F\u002Fgithub.com\u002Femyann\u002Fnemia\u002Fblob\u002Fmain\u002Fpackages\u002Ftemplates\u002Fsdlc\u002Fskills\u002Fwrite-a-spec\u002Frules.json",[53,60,67,74,81,88],{"skillNo":54,"name":55,"nameZh":6,"description":56,"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},"01M7BGEY7EKSRDFAB7B9567QKJ","groom-the-inbox","Read each card in play in the inbox against the repository and say, per card, done (the commit on the base that landed it), replaced (the decision or card that did), stale (each fact that moved, what it said and what it says now) or standing, reading closely only what the period touched. For the janitor's step, which proposes and never acts: a person removes or amends each card by the command the check writes.","https:\u002F\u002Fgithub.com\u002Femyann\u002Fnemia\u002Fblob\u002Fmain\u002Fpackages\u002Ftemplates\u002Fsdlc\u002Fskills\u002Fgroom-the-inbox\u002FSKILL.md",[],[],{"skillNo":61,"name":62,"nameZh":6,"description":63,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":64,"categories":65,"keywords":66,"publishedAt":19},"01M7B4A88P2FKWCS1RND507WHE","review-a-change","Review a change in this repository as a second engineer would: against its intent and spec, by reading the diff and the suite's verdict, with findings that name the file, the line and what goes wrong. For a read-only step that decides whether the work goes round again.","https:\u002F\u002Fgithub.com\u002Femyann\u002Fnemia\u002Fblob\u002Fmain\u002Fpackages\u002Ftemplates\u002Fsdlc\u002Fskills\u002Freview-a-change\u002FSKILL.md",[],[],{"skillNo":68,"name":69,"nameZh":6,"description":70,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":71,"categories":72,"keywords":73,"publishedAt":19},"01M7B4A2HFW8F53E390A2KM0R8","read-the-stats","Read a factory's stats since the janitor's last run, against the same length of time before it, write a short report, and propose one correction per card — an allow or deny by name, a line in an agent's body or a skill, or a refusal kept on purpose — each with the numbers behind it, never one already in the inbox. For the janitor's step, which proposes and never acts.","https:\u002F\u002Fgithub.com\u002Femyann\u002Fnemia\u002Fblob\u002Fmain\u002Fpackages\u002Ftemplates\u002Fsdlc\u002Fskills\u002Fread-the-stats\u002FSKILL.md",[],[],{"skillNo":75,"name":76,"nameZh":6,"description":77,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":78,"categories":79,"keywords":80,"publishedAt":19},"01M7B49Z4GJ2G8T76X744YHAAN","land-a-branch","Land a run's result on the base branch in this repository's own checkout, the way the step says it lands: make that one act, read its result, and write the message that records what landed — or change nothing and say why it refused. For a gated ending step (workspace: base).","https:\u002F\u002Fgithub.com\u002Femyann\u002Fnemia\u002Fblob\u002Fmain\u002Fpackages\u002Ftemplates\u002Fsdlc\u002Fskills\u002Fland-a-branch\u002FSKILL.md",[],[],{"skillNo":82,"name":83,"nameZh":6,"description":84,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":85,"categories":86,"keywords":87,"publishedAt":19},"01M7B49TMM0QAWF5WQ0QBXM9BG","implement-a-change","Build a change from its spec in this repository: tests first where a rule is claimed, small commits, the tests of what changed green for the right reason, and each review finding answered in the next round. For a step that writes the code.","https:\u002F\u002Fgithub.com\u002Femyann\u002Fnemia\u002Fblob\u002Fmain\u002Fpackages\u002Ftemplates\u002Fsdlc\u002Fskills\u002Fimplement-a-change\u002FSKILL.md",[],[],{"skillNo":89,"name":90,"nameZh":6,"description":91,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":92,"categories":93,"keywords":94,"publishedAt":19},"01M7B49PY3NQR84QJAYQZMPZG6","trial-merge-a-branch","Stages the merge of a run's result branch in the repository's own checkout, and writes its commit message.","https:\u002F\u002Fgithub.com\u002Femyann\u002Fnemia\u002Fblob\u002Fmain\u002Fpackages\u002Fagent-orchestrator\u002Ftest\u002Ffixtures\u002Fteam\u002Fskills\u002Ftrial-merge-a-branch\u002FSKILL.md",[],[]]