[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6R613TX46BB9NRZDNA1AAPE":3,"skill-same:01M6R613TX46BB9NRZDNA1AAPE":67},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":15,"categories":16,"keywords":22,"publishedAt":36,"linkOnly":37,"markdownContent":38,"markdownBody":39,"frontMatter":40,"refName":41,"commitSha":42,"skillDirectory":43,"forksCount":13,"repositoryHtmlUrl":44,"repositoryHomepageUrl":6,"ownerHtmlUrl":45,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":46,"install":47,"resources":60},"01M6R613TX46BB9NRZDNA1AAPE","write-a-skill",null,"Create new agent skills with proper structure, progressive disclosure, and bundled resources. Use when user wants to create, write, or build a new skill.","r3dlex\u002Fskills","r3dlex","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F291373160?v=4","01M6PEFNGGBVB7SHAPWXFFX933","01M6R5SDWFHT94734KJFMYE1SX",0,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Fr3dlex\u002Fskills\u002Fblob\u002Fmain\u002F03-configure-generate\u002Fwrite-a-skill\u002FSKILL.md",[17],{"categoryNo":18,"categoryName":19,"taxonomyType":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000023","Skill与智能体开发","SCENARIO","使用场景",[23,24,25,26,27,28,29,30,31,32,33,34,35],"write a skill","write","create","user","use","build","new skill","structure","new agent","resources","with","progressive disclosure","when","2026-11-10 10:12:30",false,"---\nname: write-a-skill\ndescription: Create new agent skills with proper structure, progressive disclosure, and bundled resources. Use when user wants to create, write, or build a new skill.\n---\n\n# Writing Skills\n\n## Quick Start\n\nCreate the smallest useful `SKILL.md` first, then split details into references only when needed. Read `REFERENCE.md` when you need templates, description examples, script rules, or review checklists.\n\n## Process\n\n1. **Gather requirements** — identify the task\u002Fdomain, concrete trigger phrases, use cases, required scripts, and reference material.\n2. **Draft Layer 1** — write frontmatter with `name` and a trigger-focused `description` ending in `Use when ...`.\n3. **Draft Layer 2** — write the minimal workflow in `SKILL.md`: Quick Start first, numbered steps, and only essential options.\n4. **Split Layer 3** — move templates, examples, schemas, provider variants, and advanced behavior into content-specific files.\n5. **Add scripts when useful** — use scripts for deterministic validation, formatting, repeated transforms, or precise error handling.\n6. **Review with the user** — check coverage, missing cases, unclear steps, and whether any section is too detailed or too thin.\n7. **Validate** — target \u003C=100 body lines; use a reviewed exception only when splitting cannot preserve the workflow, and never exceed 180.\n\n## Structure\n\n```text\nskill-name\u002F\n├── SKILL.md\n├── REFERENCE.md\n├── EXAMPLES.md\n└── scripts\u002F\n```\n\n## Rules\n\n- Target Layer 2 at \u003C=100 body lines. Exceptions require owner, reason, and expiry in `.ai\u002Fskills\u002Fbody-line-exceptions.json`; 180 is absolute.\n- Put one concern per reference file; avoid vague names like `details.md`.\n- Do not bury common workflow steps in references.\n- Do not include time-sensitive claims unless they can be maintained.\n- Prefer concrete examples over generic advice, but move long examples out of `SKILL.md`.\n\n## Description Budget\n\nThe frontmatter `description` is loaded into every agent's context, so keep it compact (Codex\u002FClaude metadata policy):\n\n- **Target:** `description` ≤ 160 characters. Spend the budget on concrete trigger phrases, not prose.\n- **Exceptional maximum:** 180 characters with owner, reason, and expiry; >180 always fails.\n- **Progressive disclosure first:** if a description is growing past target, the detail belongs in the body or a reference file, not the metadata.\n- **Audited exception:** only when routing clarity requires 161–180 characters, add the skill to `.ai\u002Fskills\u002Fdescription-exceptions.json`.\n\n## Codex Parity\n\nSkills must work tool-agnostically (Claude Code and Codex). Do not hard-depend on Claude\u002FOMC-only invocations: `AskUserQuestion`, `Task(subagent_type=...)`, `Skill(...)`, `subagent_type:`, `TodoWrite`, or `mcp__*` tool calls. Prefer tool-agnostic prose — for example, ask the user a question in plain markdown instead of calling the interactive question tool. `scripts\u002Fcheck-codex-parity.sh` enforces this; it scans only real invocations and ignores documented mentions inside backticks or fenced code blocks.\n\nWhen a Claude-only construct is genuinely needed, annotate it with the graceful-degradation marker and describe a plain-markdown fallback adjacent to it:\n\n```markdown\n\u003C!-- codex:optional -->\nUse AskUserQuestion to pick a branch.\nFallback (Codex \u002F plain markdown): list the options as a numbered list and ask\nthe user to reply with a number.\n```\n\nThe marker on the construct line (or the line directly above it, with no blank line between) permits that single annotated occurrence; everything else must be unmarked and tool-agnostic.\n\n`scripts\u002Fcheck-opencode-parity.sh` enforces the same bar for the opencode surface with its own denylist (Claude\u002FOMC constructs, `~\u002F.claude` and `.claude-plugin` paths, `claude` CLI invocations) and the marker `\u003C!-- opencode:optional -->`; the exemption rules are identical.\n","# Writing Skills\n\n## Quick Start\n\nCreate the smallest useful `SKILL.md` first, then split details into references only when needed. Read `REFERENCE.md` when you need templates, description examples, script rules, or review checklists.\n\n## Process\n\n1. **Gather requirements** — identify the task\u002Fdomain, concrete trigger phrases, use cases, required scripts, and reference material.\n2. **Draft Layer 1** — write frontmatter with `name` and a trigger-focused `description` ending in `Use when ...`.\n3. **Draft Layer 2** — write the minimal workflow in `SKILL.md`: Quick Start first, numbered steps, and only essential options.\n4. **Split Layer 3** — move templates, examples, schemas, provider variants, and advanced behavior into content-specific files.\n5. **Add scripts when useful** — use scripts for deterministic validation, formatting, repeated transforms, or precise error handling.\n6. **Review with the user** — check coverage, missing cases, unclear steps, and whether any section is too detailed or too thin.\n7. **Validate** — target \u003C=100 body lines; use a reviewed exception only when splitting cannot preserve the workflow, and never exceed 180.\n\n## Structure\n\n```text\nskill-name\u002F\n├── SKILL.md\n├── REFERENCE.md\n├── EXAMPLES.md\n└── scripts\u002F\n```\n\n## Rules\n\n- Target Layer 2 at \u003C=100 body lines. Exceptions require owner, reason, and expiry in `.ai\u002Fskills\u002Fbody-line-exceptions.json`; 180 is absolute.\n- Put one concern per reference file; avoid vague names like `details.md`.\n- Do not bury common workflow steps in references.\n- Do not include time-sensitive claims unless they can be maintained.\n- Prefer concrete examples over generic advice, but move long examples out of `SKILL.md`.\n\n## Description Budget\n\nThe frontmatter `description` is loaded into every agent's context, so keep it compact (Codex\u002FClaude metadata policy):\n\n- **Target:** `description` ≤ 160 characters. Spend the budget on concrete trigger phrases, not prose.\n- **Exceptional maximum:** 180 characters with owner, reason, and expiry; >180 always fails.\n- **Progressive disclosure first:** if a description is growing past target, the detail belongs in the body or a reference file, not the metadata.\n- **Audited exception:** only when routing clarity requires 161–180 characters, add the skill to `.ai\u002Fskills\u002Fdescription-exceptions.json`.\n\n## Codex Parity\n\nSkills must work tool-agnostically (Claude Code and Codex). Do not hard-depend on Claude\u002FOMC-only invocations: `AskUserQuestion`, `Task(subagent_type=...)`, `Skill(...)`, `subagent_type:`, `TodoWrite`, or `mcp__*` tool calls. Prefer tool-agnostic prose — for example, ask the user a question in plain markdown instead of calling the interactive question tool. `scripts\u002Fcheck-codex-parity.sh` enforces this; it scans only real invocations and ignores documented mentions inside backticks or fenced code blocks.\n\nWhen a Claude-only construct is genuinely needed, annotate it with the graceful-degradation marker and describe a plain-markdown fallback adjacent to it:\n\n```markdown\n\u003C!-- codex:optional -->\nUse AskUserQuestion to pick a branch.\nFallback (Codex \u002F plain markdown): list the options as a numbered list and ask\nthe user to reply with a number.\n```\n\nThe marker on the construct line (or the line directly above it, with no blank line between) permits that single annotated occurrence; everything else must be unmarked and tool-agnostic.\n\n`scripts\u002Fcheck-opencode-parity.sh` enforces the same bar for the opencode surface with its own denylist (Claude\u002FOMC constructs, `~\u002F.claude` and `.claude-plugin` paths, `claude` CLI invocations) and the marker `\u003C!-- opencode:optional -->`; the exemption rules are identical.\n",{"name":5,"description":7},"main","037fe34299eb9c4a30b51f1a6bd101eae1c469b1","03-configure-generate\u002Fwrite-a-skill","https:\u002F\u002Fgithub.com\u002Fr3dlex\u002Fskills","https:\u002F\u002Fgithub.com\u002Fr3dlex","2026-10-29 16:22:54",{"source":8,"skillName":5,"commands":48,"prompt":58,"zipAvailable":59},[49,52,55],{"tool":50,"command":51},"npx","npx skills add r3dlex\u002Fskills --skill write-a-skill",{"tool":53,"command":54},"bunx","bunx skills add r3dlex\u002Fskills --skill write-a-skill",{"tool":56,"command":57},"pnpm","pnpm dlx skills add r3dlex\u002Fskills --skill write-a-skill","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fr3dlex\u002Fskills\u002Fblob\u002Fmain\u002F03-configure-generate\u002Fwrite-a-skill\u002FSKILL.md",true,[61],{"path":62,"resourceType":63,"resourceTypeDesc":64,"sizeBytes":65,"sourceHtmlUrl":66},"REFERENCE.md","REFERENCE","参考文档",1896,"https:\u002F\u002Fgithub.com\u002Fr3dlex\u002Fskills\u002Fblob\u002Fmain\u002F03-configure-generate\u002Fwrite-a-skill\u002FREFERENCE.md",[68,76,85,94,104,127],{"skillNo":69,"name":70,"nameZh":6,"description":71,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":72,"categories":73,"keywords":74,"publishedAt":36},"01M6R652E42RZT7M2V3RWW991Z","skill","Fixture.","https:\u002F\u002Fgithub.com\u002Fr3dlex\u002Fskills\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fcatalog-malformed\u002Fskill\u002FSKILL.md",[],[75],"fixture",{"skillNo":77,"name":78,"nameZh":6,"description":79,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":80,"categories":81,"keywords":83,"publishedAt":36},"01M6R64MTPF0FGWHFS4DWP40GR","stable-skill","Fixture stable-skill.","https:\u002F\u002Fgithub.com\u002Fr3dlex\u002Fskills\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fcatalog-lifecycle\u002Fstable-skill\u002FSKILL.md",[82],{"categoryNo":18,"categoryName":19,"taxonomyType":20,"taxonomyTypeDesc":21},[84,75],"stable skill",{"skillNo":86,"name":87,"nameZh":6,"description":88,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":89,"categories":90,"keywords":92,"publishedAt":36},"01M6R64HD4QVAYM2R0JSYNVDJW","experimental-skill","Fixture experimental-skill.","https:\u002F\u002Fgithub.com\u002Fr3dlex\u002Fskills\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fcatalog-lifecycle\u002Fexperimental-skill\u002FSKILL.md",[91],{"categoryNo":18,"categoryName":19,"taxonomyType":20,"taxonomyTypeDesc":21},[93,75],"experimental skill",{"skillNo":95,"name":96,"nameZh":6,"description":97,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":98,"categories":99,"keywords":101,"publishedAt":36},"01M6R64FZH05TEJQW75QX1E5XD","deprecated-skill","Fixture deprecated-skill.","https:\u002F\u002Fgithub.com\u002Fr3dlex\u002Fskills\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fcatalog-lifecycle\u002Fdeprecated-skill\u002FSKILL.md",[100],{"categoryNo":18,"categoryName":19,"taxonomyType":20,"taxonomyTypeDesc":21},[102,103,75],"deprecated skill","deprecated",{"skillNo":105,"name":106,"nameZh":6,"description":107,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":108,"categories":109,"keywords":110,"publishedAt":36},"01M6R64D13X5DC980GREWFSSFQ","karpathy-guidelines","Apply think-before-coding, simplicity, surgical-change, and goal-driven engineering rules. Use when planning, implementing, or reviewing code.","https:\u002F\u002Fgithub.com\u002Fr3dlex\u002Fskills\u002Fblob\u002Fmain\u002Freference\u002Fgolden-skills\u002F.agents\u002Fskills\u002Fkarpathy-guidelines\u002FSKILL.md",[],[111,112,113,114,27,115,116,117,118,119,120,121,35,122,123,124,125,126],"karpathy guidelines","karpathy","guidelines","Rules","simplicity","goal","think","driven","reviewing code","implementing","apply","coding","change","code","planning","engineering",{"skillNo":128,"name":106,"nameZh":6,"description":107,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":129,"categories":130,"keywords":131,"publishedAt":36},"01M6R640ZS60BRV9P49FT46ZJX","https:\u002F\u002Fgithub.com\u002Fr3dlex\u002Fskills\u002Fblob\u002Fmain\u002Freference\u002Fgolden-root\u002F.agents\u002Fskills\u002Fkarpathy-guidelines\u002FSKILL.md",[],[113,111,112,123,118,124,125,126,122,114,27,115,116,117,119,120,121,35]]