[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-2053":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":11,"languages":10,"totalLinesOfCode":10,"stars":12,"forks":13,"watchers":14,"openIssues":15,"contributorsCount":15,"subscribersCount":15,"size":15,"stars1d":15,"stars7d":15,"stars30d":16,"stars90d":15,"forks30d":15,"starsTrendScore":15,"compositeScore":17,"rankGlobal":10,"rankLanguage":10,"license":18,"archived":19,"fork":19,"defaultBranch":20,"hasWiki":19,"hasPages":19,"topics":21,"createdAt":10,"pushedAt":10,"updatedAt":28,"readmeContent":29,"aiSummary":30,"trendingCount":15,"starSnapshotCount":15,"syncStatus":16,"lastSyncTime":31,"discoverSource":32},2053,"skillplus","eight-acres-lab\u002Fskillplus","eight-acres-lab","A compilable skill package standard for reliable content-generation agents.","https:\u002F\u002Fpointeight.ai",null,"TypeScript",306,20,6,0,2,3.97,"MIT License",false,"main",[22,23,24,25,26,27],"ai-agent","compiler","content-generation","llm","prompt-engineering","skill-package","2026-06-12 02:00:36","# Skill-Plus\n\nCompilable skill packages for content-generation agents.\n\n```bash\nnpm i -g @e8s\u002Fskillplus\n\nskillplus list                          # browse the official catalog\nskillplus search \"phone snapshot food\"  # find a skill by keyword\nskillplus show food-street-realism      # see metadata\nskillplus food-street-realism --target skill-md > ~\u002F.claude\u002Fskills\u002Ffood.md\n```\n\nA skill is a directory of YAML + Markdown that the compiler turns into one of several target formats: structured JSON for the [OpenMelon](https:\u002F\u002Fgithub.com\u002Feight-acres-lab\u002Fopenmelon) runtime, a portable `skill.md` for Claude Code \u002F Cursor \u002F any agent, vendor-specific prompt bundles, evaluation checklists, or provenance templates. The compiler is deterministic and never calls a model — that's the runtime's job.\n\n## Install\n\n```bash\nnpm i -g @e8s\u002Fskillplus\n```\n\nRequires Node ≥20.\n\n## Commands\n\n```\nskillplus \u003Cid-or-path> [--target ...] [--model-profile ...] [--locale ...]   compile\nskillplus list [--tag \u003Ct>] [--mine] [--json]                                 list available skills\nskillplus search \u003Cquery> [--json]                                            fuzzy search\nskillplus show \u003Cid> [--json]                                                 print metadata\nskillplus copy \u003Cid> [--force]                                                fork a bundled skill into ~\u002F.skillplus\u002Flocal\u002F\nskillplus init \u003Cid> [--from \u003Cexisting>]                                      scaffold a new skill\nskillplus update [--check] [--dry-run]                                       self-update via npm\n```\n\n`skillplus \u003Ccmd> --help` for flags.\n\n### Compile\n\n```bash\n# By skill id (resolves against ~\u002F.skillplus\u002Flocal\u002F first, then bundled)\nskillplus food-street-realism --target openmelon --locale zh-CN\n\n# By path (local development)\nskillplus .\u002Fmy-package.skillplus --target skill-md\n\n# Drop-in for Claude Code\nskillplus food-street-realism --target skill-md > ~\u002F.claude\u002Fskills\u002Ffood.md\n```\n\nTargets: `openmelon`, `skill-md`, `prompt-bundle`, `eval`, `provenance`.\n\n### Browse the catalog\n\n`skillplus list` shows everything available — bundled official skills plus your own under `~\u002F.skillplus\u002Flocal\u002F`. Local takes precedence in resolution: `skillplus copy food-street-realism` makes a local fork, which then shadows the official version.\n\n```\nID                   VERSION  SOURCE    DESCRIPTION\n---------------------------------------------------\nbrand-logo           0.1.0    bundled   Single-mark brand logos for OSS, dev tools…\nfood-street-realism  0.1.0    bundled   Real-feeling phone-snapshot food posts…\n```\n\n### Update\n\n```bash\nskillplus update --check     # see if a newer version exists\nskillplus update             # install it\n```\n\nThis runs `npm install -g @e8s\u002Fskillplus@latest` under the hood. Mirrors `claude update`.\n\n## Featured skills\n\nCurated picks from the catalog (full list: [`skills\u002FREADME.md`](skills\u002FREADME.md) or `skillplus list`).\n\n| Skill | What it does |\n|---|---|\n| [`food-street-realism`](skills\u002Ffood-street-realism.skillplus\u002F) | Real-feeling phone-snapshot food posts for 探店 \u002F Xiaohongshu |\n| [`brand-logo`](skills\u002Fbrand-logo.skillplus\u002F) | Flat-vector brand logos that read at favicon size |\n\n## Contribute\n\nSkill-Plus grows with the community. New skills are accepted via PR.\n\nThe contribution path is short:\n\n```bash\nskillplus init my-skill          # scaffold .\u002Fmy-skill.skillplus\u002F\n# fill in the TODOs (skillplus.yaml, .search, prompts\u002F, schema\u002F, eval\u002F)\nskillplus my-skill --target openmelon --locale en   # compile to verify\n# open a PR moving .\u002Fmy-skill.skillplus\u002F into skills\u002F\n```\n\nRequired for acceptance:\n\n- `.search` file with a description (≤200 chars) and 1-10 kebab-case tags\n- Manifest fields per [SPEC.md](SPEC.md)\n- At least one locale, one model profile, one base prompt\n- Evaluation checklist + failure modes\n- A real example input\n- Demonstrably useful (not a \"tweak the temperature\" reskin of an existing skill)\n\nSee [CONTRIBUTING.md](CONTRIBUTING.md) for the full submission requirements + featured-tier criteria.\n\n## Public API\n\n```ts\nimport { compilePackage, renderTarget, CompileError } from \"@e8s\u002Fskillplus\"\n\nconst compiled = compilePackage({\n  packageDir: \".\u002Ffood-street-realism.skillplus\",\n  target: \"openmelon\",\n  modelProfile: \"gpt-image-family\",\n  locale: \"zh-CN\",\n  varsOverride: { realism_level: \"high\" },\n})\nconst { text } = renderTarget(compiled)\n```\n\n## Vendor model profiles\n\nPer-vendor prompt-style preferences and gotchas live in [`model-profiles\u002F`](model-profiles\u002F). Today: Anthropic Claude. More coming.\n\n## License\n\n[MIT](LICENSE).\n","Skill-Plus 是一个用于内容生成代理的可编译技能包标准。它通过将YAML和Markdown格式的技能文件编译成多种目标格式（如JSON、Markdown等），以适应不同的运行时环境或模型需求，支持OpenMelon运行时、Claude Code等平台。项目采用TypeScript开发，提供命令行工具便于技能包的浏览、搜索、显示及编译操作。适合需要定制化内容生成能力的应用场景，例如自动化社交媒体内容创作、基于特定主题的信息生成等，尤其对于开发者来说是一个强大的工具，能够简化AI辅助内容创建流程。","2026-05-31 02:45:11","CREATED_QUERY"]