[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7AY5NPXYDZS06MNV464CV2H":3,"skill-same:01M7AY5NPXYDZS06MNV464CV2H":46},{"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":25,"commitSha":26,"skillDirectory":27,"forksCount":14,"repositoryHtmlUrl":28,"repositoryHomepageUrl":6,"ownerHtmlUrl":29,"ownerBlogUrl":6,"ownerDescription":30,"updatedAt":31,"install":32,"resources":45},"01M7AY5NPXYDZS06MNV464CV2H","adadr",null,"analyze, design, adversarial-review, decision-gate, realize","lost-rob0t\u002Fskills","lost-rob0t","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F104283403?v=4","01M7AE2T5GZHE2CF6ANP6TKG0E","nsaspy","01M7AY4Z7QRGZRBX43TN2XT0GR",0,"BSD-3-Clause","https:\u002F\u002Fgithub.com\u002Flost-rob0t\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fadadr\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: adadr\ndescription: analyze, design, adversarial-review, decision-gate, realize\ncompatibility: Agent Skills-compatible software engineering agent with repository access, evidence gathering, and project approval metadata\n---\n\n# ADADR\n\nEvidence-driven software architecture loop.\n\n## Control-plane discovery gate\n\nBefore architecture work, discover the project's canonical location or workflow for research, design, approval state, and implementation promotion. Use repository instructions, project configuration, or current authoritative project state.\n\nDo not invent a new `research\u002F`, `design\u002F`, `adadr\u002F`, or approval directory just to satisfy this skill.\n\nIf no canonical location\u002Fworkflow is defined, or the defined location is not actually used by the project, stop before Analyze and ask the user to choose one mode:\n\n1. **ADADR** — human-gated software design. Research approval and design approval are explicit user gates.\n2. **Auto-RAGE** — the coding worker runs the complete ADADR software-design loop itself and records the research, design, review, decision, and verification evidence without stopping for user approval.\n\nPersist the selected mode in durable project instructions\u002Fconfiguration when authorized so future workers do not ask again. Never silently choose Auto-RAGE.\n\n## Loop\n\n### A — Analyze\n\nEstablish the real problem before proposing architecture. Inspect current code\u002Ftests, existing research\u002Fdesign, recent changes, runtime evidence, relevant specifications, and serious alternatives. Separate verified facts, inference, contradictions, and unknowns. For defects, prefer a deterministic failing regression as primary evidence.\n\n### D — Design\n\nDerive the smallest coherent design from the evidence. Define ownership and state boundaries, invariants, lifecycle, failure\u002Frecovery behavior, concurrency and resource bounds, security\u002Fauthority boundaries, compatibility\u002Fmigration constraints, acceptance criteria, and exact verification gates. Record rejected alternatives and why they lose.\n\n### A — Adversarial review\n\nAttack the design before coding. Construct counterexamples and inspect race, stale-state, partial-failure, cancellation, restart, security, provenance, compatibility, and operational failure cases. If the design fails, revise it and repeat this phase.\n\n### D — Decision gate\n\nIn **ADADR mode**, use the project's real human approval state. Research approval authorizes design work only; design approval authorizes implementation within that scope only. Missing approval blocks implementation.\n\nIn **Auto-RAGE mode**, the worker may record its own software-design decision after adversarial review, but it must still produce the explicit decision record before implementation.\n\nNever infer human approval from a merge, green CI, an existing file, or model confidence.\n\n### R — Realize\n\nImplement only the approved\u002Frecorded design using the repository's required implementation workflow. Keep changes traceable to design acceptance criteria. If implementation evidence disproves the design, preserve the evidence and return to Analyze rather than silently drifting architecture.\n\n## Human-gated progression\n\n`research -> explicit research approval -> RAGE produces design -> explicit design approval -> implementation`\n\n## Auto-RAGE progression\n\n`research -> design -> adversarial review -> recorded worker decision -> implementation`\n\n## RAGE integration\n\nWhen invoked by RAGE, ADADR owns control-plane discovery and architecture reasoning. RAGE owns queue selection, run identity, TDD discipline, exact-head verification, pull-request handling, merge gates, and backlog continuation.\n\n## Rules\n\n- Repository instructions and current evidence outrank remembered architecture.\n- Bug-first\u002FTDD-first repository policy may establish evidence before Analyze.\n- Do not invent an ADADR directory\u002Fcontrol plane when none exists.\n- Do not silently choose Auto-RAGE.\n- Do not hide architecture changes inside implementation patches.\n- Prefer canonical project research\u002Fdesign artifacts when the repository defines them as authoritative.\n","# ADADR\n\nEvidence-driven software architecture loop.\n\n## Control-plane discovery gate\n\nBefore architecture work, discover the project's canonical location or workflow for research, design, approval state, and implementation promotion. Use repository instructions, project configuration, or current authoritative project state.\n\nDo not invent a new `research\u002F`, `design\u002F`, `adadr\u002F`, or approval directory just to satisfy this skill.\n\nIf no canonical location\u002Fworkflow is defined, or the defined location is not actually used by the project, stop before Analyze and ask the user to choose one mode:\n\n1. **ADADR** — human-gated software design. Research approval and design approval are explicit user gates.\n2. **Auto-RAGE** — the coding worker runs the complete ADADR software-design loop itself and records the research, design, review, decision, and verification evidence without stopping for user approval.\n\nPersist the selected mode in durable project instructions\u002Fconfiguration when authorized so future workers do not ask again. Never silently choose Auto-RAGE.\n\n## Loop\n\n### A — Analyze\n\nEstablish the real problem before proposing architecture. Inspect current code\u002Ftests, existing research\u002Fdesign, recent changes, runtime evidence, relevant specifications, and serious alternatives. Separate verified facts, inference, contradictions, and unknowns. For defects, prefer a deterministic failing regression as primary evidence.\n\n### D — Design\n\nDerive the smallest coherent design from the evidence. Define ownership and state boundaries, invariants, lifecycle, failure\u002Frecovery behavior, concurrency and resource bounds, security\u002Fauthority boundaries, compatibility\u002Fmigration constraints, acceptance criteria, and exact verification gates. Record rejected alternatives and why they lose.\n\n### A — Adversarial review\n\nAttack the design before coding. Construct counterexamples and inspect race, stale-state, partial-failure, cancellation, restart, security, provenance, compatibility, and operational failure cases. If the design fails, revise it and repeat this phase.\n\n### D — Decision gate\n\nIn **ADADR mode**, use the project's real human approval state. Research approval authorizes design work only; design approval authorizes implementation within that scope only. Missing approval blocks implementation.\n\nIn **Auto-RAGE mode**, the worker may record its own software-design decision after adversarial review, but it must still produce the explicit decision record before implementation.\n\nNever infer human approval from a merge, green CI, an existing file, or model confidence.\n\n### R — Realize\n\nImplement only the approved\u002Frecorded design using the repository's required implementation workflow. Keep changes traceable to design acceptance criteria. If implementation evidence disproves the design, preserve the evidence and return to Analyze rather than silently drifting architecture.\n\n## Human-gated progression\n\n`research -> explicit research approval -> RAGE produces design -> explicit design approval -> implementation`\n\n## Auto-RAGE progression\n\n`research -> design -> adversarial review -> recorded worker decision -> implementation`\n\n## RAGE integration\n\nWhen invoked by RAGE, ADADR owns control-plane discovery and architecture reasoning. RAGE owns queue selection, run identity, TDD discipline, exact-head verification, pull-request handling, merge gates, and backlog continuation.\n\n## Rules\n\n- Repository instructions and current evidence outrank remembered architecture.\n- Bug-first\u002FTDD-first repository policy may establish evidence before Analyze.\n- Do not invent an ADADR directory\u002Fcontrol plane when none exists.\n- Do not silently choose Auto-RAGE.\n- Do not hide architecture changes inside implementation patches.\n- Prefer canonical project research\u002Fdesign artifacts when the repository defines them as authoritative.\n",{"name":5,"description":7,"compatibility":24},"Agent Skills-compatible software engineering agent with repository access, evidence gathering, and project approval metadata","main","4b51593eee1b59ce94f1a7bb84b9a1393414906a","skills\u002Fadadr","https:\u002F\u002Fgithub.com\u002Flost-rob0t\u002Fskills","https:\u002F\u002Fgithub.com\u002Flost-rob0t","HackerMan\r\nLISP\u002Fnim + nixos\r\n\r\nMonero: \r\n8Bk9yV92PqAepeeCHvWVXsLwnpMWMmB99XoGgLaxWEZNfDXXcfzDWyH56XaGs6K5YFZckaPjQGsPXdRZDokgEFbKSUG9H","2026-11-14 09:27:40",{"source":8,"skillName":5,"commands":33,"prompt":43,"zipAvailable":44},[34,37,40],{"tool":35,"command":36},"npx","npx skills add lost-rob0t\u002Fskills --skill adadr",{"tool":38,"command":39},"bunx","bunx skills add lost-rob0t\u002Fskills --skill adadr",{"tool":41,"command":42},"pnpm","pnpm dlx skills add lost-rob0t\u002Fskills --skill adadr","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Flost-rob0t\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fadadr\u002FSKILL.md",true,[],[47,54,61,68,75,82],{"skillNo":48,"name":49,"nameZh":6,"description":50,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":51,"categories":52,"keywords":53,"publishedAt":19},"01M7AYCQH81M6KWJF8SXDK3C0C","unknown-frontmatter","validator, fixture","https:\u002F\u002Fgithub.com\u002Flost-rob0t\u002Fskills\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Funknown-frontmatter\u002FSKILL.md",[],[],{"skillNo":55,"name":56,"nameZh":6,"description":57,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":58,"categories":59,"keywords":60,"publishedAt":19},"01M7AYBFAATN244PFVY74JWCWM","zara-phone-pair","zara, phone, pairing, adb, wireless-debugging, curve, enrollment, android","https:\u002F\u002Fgithub.com\u002Flost-rob0t\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fzara-phone-pair\u002FSKILL.md",[],[],{"skillNo":62,"name":63,"nameZh":6,"description":64,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":65,"categories":66,"keywords":67,"publishedAt":19},"01M7AYB9NY8AX86EN0A1PSVBFK","zara-mcp","zara, mcp, stdio, http, tools, resources, prompts, debugging","https:\u002F\u002Fgithub.com\u002Flost-rob0t\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fzara-mcp\u002FSKILL.md",[],[],{"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},"01M7AYB63ZZTN44G20M900X2WZ","youtube-context","youtube, transcripts, yt-dlp, whisper, video-context","https:\u002F\u002Fgithub.com\u002Flost-rob0t\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fyoutube-context\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},"01M7AYB3JEZGMP6YQ91X7BYYED","worker-orchestration","workers, routing, delegation, fleet, budgets, concurrency","https:\u002F\u002Fgithub.com\u002Flost-rob0t\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fworker-orchestration\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},"01M7AYAZZCC7TXT978GBHEPV47","task-steward-worker","steward, rage, leases, fencing, heartbeat, receipts","https:\u002F\u002Fgithub.com\u002Flost-rob0t\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Ftask-steward-worker\u002FSKILL.md",[],[]]