[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M77Z4DNAPPQKQ2KSNJ0JBXT8":3,"skill-same:01M77Z4DNAPPQKQ2KSNJ0JBXT8":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":24,"commitSha":25,"skillDirectory":26,"forksCount":27,"repositoryHtmlUrl":28,"repositoryHomepageUrl":6,"ownerHtmlUrl":29,"ownerBlogUrl":30,"ownerDescription":31,"updatedAt":32,"install":33,"resources":46},"01M77Z4DNAPPQKQ2KSNJ0JBXT8","efficient-fable",null,"Use when running Claude Fable on codebase-heavy or token-heavy work and the user wants Fable to orchestrate research, coding, and testing while cheaper subagents do bounded heavy lifting.","compiled-run\u002Fmarkless","compiled-run","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F312797118?v=4","01M77Z248Q9PCPK92HPHM6X174","Compiled.Run","01M77Z249DKAP6PE0Y5CNFSSMM",3,"MIT","https:\u002F\u002Fgithub.com\u002Fcompiled-run\u002Fmarkless\u002Fblob\u002Fmain\u002F.codex\u002Fskills\u002Fefficient-fable\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: efficient-fable\ndescription: Use when running Claude Fable on codebase-heavy or token-heavy work and the user wants Fable to orchestrate research, coding, and testing while cheaper subagents do bounded heavy lifting.\n---\n\n# Efficient Fable\n\nUse Claude Fable as the orchestrator, architect, synthesizer, and final judge.\nUse cheaper subagents for token-heavy research, coding, testing, and\nsummarization that do not require Fable's full judgment.\n\n## Where Fable Shines\n\nReserve Fable for:\n\n- Decomposing ambiguous work into clean parallel slices.\n- Architecture, product, and safety tradeoffs.\n- Reading conflicting subagent reports and deciding what matters.\n- Integrating partial implementations into one coherent plan.\n- Final review, risk assessment, and user-facing synthesis.\n\n## Delegation Pattern\n\n1. Name the expensive-token risk: large repo search, long logs, broad docs, or\n   repetitive edits.\n2. Split independent work into subagents before reading everything yourself.\n3. Use cheaper models for research scans, inventory, search summaries, narrow\n   bug hunts, browser\u002Ftesting passes, test output reduction, and bounded code\n   edits.\n4. Ask subagents for concise evidence: files, line references, commands run,\n   diffs, uncertainties, and stop conditions they hit.\n5. Spend Fable tokens on the decision layer: compare results, resolve conflicts,\n   choose the implementation path, and review the final patch.\n\nPrefer parallel subagents when the slices do not depend on each other. Keep\nblocking or highly coupled work local.\n\n## Handoff Packets\n\nWrite delegated prompts as if the subagent has no useful chat context. Include\nonly the context it needs:\n\n- The repo path and exact objective.\n- The files, packages, or surfaces in scope and anything explicitly out of\n  scope.\n- The evidence format to return: files, line refs, commands, diffs, failures,\n  screenshots, and uncertainty.\n- The verification commands or browser flows to run, plus what success should\n  look like when that is knowable.\n- Stop conditions: if the code does not match the prompt, a command fails after\n  a reasonable retry, or the task needs out-of-scope files, stop and report\n  instead of improvising.\n\n## Vetting Delegated Work\n\nTreat subagent reports as leads, not facts. Before using a high-impact finding,\nopening a PR, or telling the user the work is done, Fable should reopen the\nimportant cited files, confirm the relevant line refs or failures, and review\nthe final diff against the task. Let lighter agents gather signal; keep\ntruth-judgment with Fable.\n\n## Common Scenarios\n\nTreat these as soft defaults, not rigid rules:\n\n- Research: ask lighter agents to scan docs, prior art, APIs, and repo surfaces;\n  Fable decides what evidence changes the plan.\n- Coding: give cheaper agents bounded edits or candidate patches; Fable owns\n  shared-file coordination, integration, and final review.\n- Testing: have Fable suggest the validation direction and the scripts or\n  browser checks that matter. Let lighter agents run targeted tests, browser\n  flows, screenshots, and log reduction, then report exact commands, failures,\n  likely causes, and whether failures look flaky, environmental, or real.\n- Debugging: use cheaper agents to cluster logs, reproduce issues, and try\n  small fixes; Fable decides which diagnosis is most trustworthy.\n\nIf a task is tiny or the validation itself needs delicate judgment, keep it\nwith Fable.\n\n## Diagram\n\nUse `assets\u002Ffable-orchestrator.excalidraw` when a visual explanation helps.\n\n## Claims\n\nFor codebase-heavy work, it is reasonable to describe this as up to 3-5x more\ncost-efficient and 2-4x faster when independent research, coding, or testing\nslices can run in parallel. Treat those as workload-dependent estimates, not\nguarantees.\n\nGood launch copy:\n\n> Make Claude Fable more efficient by using cheaper subagents for token-heavy\n> research, coding, and testing, saving Fable for judgment, architecture,\n> synthesis, and final review.\n","# Efficient Fable\n\nUse Claude Fable as the orchestrator, architect, synthesizer, and final judge.\nUse cheaper subagents for token-heavy research, coding, testing, and\nsummarization that do not require Fable's full judgment.\n\n## Where Fable Shines\n\nReserve Fable for:\n\n- Decomposing ambiguous work into clean parallel slices.\n- Architecture, product, and safety tradeoffs.\n- Reading conflicting subagent reports and deciding what matters.\n- Integrating partial implementations into one coherent plan.\n- Final review, risk assessment, and user-facing synthesis.\n\n## Delegation Pattern\n\n1. Name the expensive-token risk: large repo search, long logs, broad docs, or\n   repetitive edits.\n2. Split independent work into subagents before reading everything yourself.\n3. Use cheaper models for research scans, inventory, search summaries, narrow\n   bug hunts, browser\u002Ftesting passes, test output reduction, and bounded code\n   edits.\n4. Ask subagents for concise evidence: files, line references, commands run,\n   diffs, uncertainties, and stop conditions they hit.\n5. Spend Fable tokens on the decision layer: compare results, resolve conflicts,\n   choose the implementation path, and review the final patch.\n\nPrefer parallel subagents when the slices do not depend on each other. Keep\nblocking or highly coupled work local.\n\n## Handoff Packets\n\nWrite delegated prompts as if the subagent has no useful chat context. Include\nonly the context it needs:\n\n- The repo path and exact objective.\n- The files, packages, or surfaces in scope and anything explicitly out of\n  scope.\n- The evidence format to return: files, line refs, commands, diffs, failures,\n  screenshots, and uncertainty.\n- The verification commands or browser flows to run, plus what success should\n  look like when that is knowable.\n- Stop conditions: if the code does not match the prompt, a command fails after\n  a reasonable retry, or the task needs out-of-scope files, stop and report\n  instead of improvising.\n\n## Vetting Delegated Work\n\nTreat subagent reports as leads, not facts. Before using a high-impact finding,\nopening a PR, or telling the user the work is done, Fable should reopen the\nimportant cited files, confirm the relevant line refs or failures, and review\nthe final diff against the task. Let lighter agents gather signal; keep\ntruth-judgment with Fable.\n\n## Common Scenarios\n\nTreat these as soft defaults, not rigid rules:\n\n- Research: ask lighter agents to scan docs, prior art, APIs, and repo surfaces;\n  Fable decides what evidence changes the plan.\n- Coding: give cheaper agents bounded edits or candidate patches; Fable owns\n  shared-file coordination, integration, and final review.\n- Testing: have Fable suggest the validation direction and the scripts or\n  browser checks that matter. Let lighter agents run targeted tests, browser\n  flows, screenshots, and log reduction, then report exact commands, failures,\n  likely causes, and whether failures look flaky, environmental, or real.\n- Debugging: use cheaper agents to cluster logs, reproduce issues, and try\n  small fixes; Fable decides which diagnosis is most trustworthy.\n\nIf a task is tiny or the validation itself needs delicate judgment, keep it\nwith Fable.\n\n## Diagram\n\nUse `assets\u002Ffable-orchestrator.excalidraw` when a visual explanation helps.\n\n## Claims\n\nFor codebase-heavy work, it is reasonable to describe this as up to 3-5x more\ncost-efficient and 2-4x faster when independent research, coding, or testing\nslices can run in parallel. Treat those as workload-dependent estimates, not\nguarantees.\n\nGood launch copy:\n\n> Make Claude Fable more efficient by using cheaper subagents for token-heavy\n> research, coding, and testing, saving Fable for judgment, architecture,\n> synthesis, and final review.\n",{"name":5,"description":7},"main","a65038eb31b69b03c9d7b57b7a8149386d02064c",".codex\u002Fskills\u002Fefficient-fable",0,"https:\u002F\u002Fgithub.com\u002Fcompiled-run\u002Fmarkless","https:\u002F\u002Fgithub.com\u002Fcompiled-run","https:\u002F\u002Fcompiled.run\u002F","Building the next generation of compiler software","2026-11-15 05:16:50",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":45},[35,38,41],{"tool":36,"command":37},"npx","npx skills add compiled-run\u002Fmarkless --skill efficient-fable",{"tool":39,"command":40},"bunx","bunx skills add compiled-run\u002Fmarkless --skill efficient-fable",{"tool":42,"command":43},"pnpm","pnpm dlx skills add compiled-run\u002Fmarkless --skill efficient-fable","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fcompiled-run\u002Fmarkless\u002Fblob\u002Fmain\u002F.codex\u002Fskills\u002Fefficient-fable\u002FSKILL.md",true,[47,53,57,61],{"path":48,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":51,"sourceHtmlUrl":52},"assets\u002Ffable-orchestrator-dark.png","ASSET","资产文件",111350,"https:\u002F\u002Fgithub.com\u002Fcompiled-run\u002Fmarkless\u002Fblob\u002Fmain\u002F.codex\u002Fskills\u002Fefficient-fable\u002Fassets\u002Ffable-orchestrator-dark.png",{"path":54,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":55,"sourceHtmlUrl":56},"assets\u002Ffable-orchestrator.excalidraw",19562,"https:\u002F\u002Fgithub.com\u002Fcompiled-run\u002Fmarkless\u002Fblob\u002Fmain\u002F.codex\u002Fskills\u002Fefficient-fable\u002Fassets\u002Ffable-orchestrator.excalidraw",{"path":58,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":59,"sourceHtmlUrl":60},"assets\u002Ffable-orchestrator.png",111227,"https:\u002F\u002Fgithub.com\u002Fcompiled-run\u002Fmarkless\u002Fblob\u002Fmain\u002F.codex\u002Fskills\u002Fefficient-fable\u002Fassets\u002Ffable-orchestrator.png",{"path":62,"resourceType":63,"resourceTypeDesc":64,"sizeBytes":65,"sourceHtmlUrl":66},"README.md","REFERENCE","参考文档",2904,"https:\u002F\u002Fgithub.com\u002Fcompiled-run\u002Fmarkless\u002Fblob\u002Fmain\u002F.codex\u002Fskills\u002Fefficient-fable\u002FREADME.md",[68,75,82,89,96,104],{"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},"01M77Z5Z2DN4WKQD805NF4DBQC","review-with-engineer-lenses","Use when reviewing API changes, API design, technical docs, developer experience, or repo changes through junior engineer, senior engineer, platform engineer, and AI-agent perspectives.","https:\u002F\u002Fgithub.com\u002Fcompiled-run\u002Fmarkless\u002Fblob\u002Fmain\u002F.ruler\u002Fskills\u002Freview-with-engineer-lenses\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},"01M77Z5XCP92P5J58JK56SGTHS","markless-spec-maintenance","Use when updating, reviewing, or reconciling Markless specifications under specs\u002F.","https:\u002F\u002Fgithub.com\u002Fcompiled-run\u002Fmarkless\u002Fblob\u002Fmain\u002F.ruler\u002Fskills\u002Fmarkless-spec-maintenance\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},"01M77Z5TXA1T7A32ECR5R3PB1S","markless-implementation","Use for Markless implementation, compiler, bundler, performance, testing, and release work.","https:\u002F\u002Fgithub.com\u002Fcompiled-run\u002Fmarkless\u002Fblob\u002Fmain\u002F.ruler\u002Fskills\u002Fmarkless-implementation\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},"01M77Z5NWFHXM8Z0EAWV3X3598","markless-component-research","Research component accessibility patterns and API design for a @markless\u002Fui family by analyzing headless UI libraries, the QDS reference, and accessibility specifications. Use before designing or migrating any component family — when asked to research a UI component, create its research.md, analyze accessibility patterns across component libraries, or document keyboard interactions and ARIA attributes.","https:\u002F\u002Fgithub.com\u002Fcompiled-run\u002Fmarkless\u002Fblob\u002Fmain\u002F.ruler\u002Fskills\u002Fmarkless-component-research\u002FSKILL.md",[],[],{"skillNo":97,"name":98,"nameZh":6,"description":99,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":100,"sourceHtmlUrl":101,"categories":102,"keywords":103,"publishedAt":19},"01M77Z5HFPMKEZYB1MWWNRFSND","markless-anatomy","Build a live pen-and-ink figure for a markless.dev docs page, one idea drawn as 2 or 3 objects where the reader acts once and sees one change, in the same hand and ink as the State figure. Use whenever you add or rewrite a markless.dev docs page (website\u002Fpages\u002F**), explain a Markless concept visually, add a figure or diagram to the docs, or someone mentions \"anatomy\", the wired drawing, or website\u002Fcomponents\u002Ffigures\u002F.","https:\u002F\u002Fgithub.com\u002Fcompiled-run\u002Fmarkless\u002Fblob\u002Fmain\u002F.ruler\u002Fskills\u002Fmarkless-anatomy\u002FLICENSE-UPSTREAM","https:\u002F\u002Fgithub.com\u002Fcompiled-run\u002Fmarkless\u002Fblob\u002Fmain\u002F.ruler\u002Fskills\u002Fmarkless-anatomy\u002FSKILL.md",[],[],{"skillNo":105,"name":106,"nameZh":6,"description":107,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":108,"categories":109,"keywords":110,"publishedAt":19},"01M77Z5D0FTA3XSEVNVNX2P7HV","guessless","Prove structural claims about JavaScript\u002FTypeScript with a guessless receipt. Use when about to say \"all call sites\", \"every reference\", \"nothing else imports this\", or \"safe to delete\"; when renaming or deleting an exported symbol; when a claim gate has blocked a turn asking for a receipt; or when auditing whether a change is complete. Do not use for non-JS\u002FTS languages.","https:\u002F\u002Fgithub.com\u002Fcompiled-run\u002Fmarkless\u002Fblob\u002Fmain\u002F.ruler\u002Fskills\u002Fguessless\u002FSKILL.md",[],[]]