[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M79AGTDB3VE9MKBXN7ZXKPYM":3,"skill-same:01M79AGTDB3VE9MKBXN7ZXKPYM":47},{"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":29,"ownerHtmlUrl":30,"ownerBlogUrl":6,"ownerDescription":31,"updatedAt":32,"install":33,"resources":46},"01M79AGTDB3VE9MKBXN7ZXKPYM","zirkel-explore-branches",null,"Explores multiple genuinely-distinct approaches to a scoped problem in parallel, scores each on Feasibility\u002FImpact\u002FRisk, and selects a winner — so the work does not anchor on the first idea. Use when a task has more than one viable approach and picking the first that comes to mind would be a mistake: \"what are our options here\", \"don't just do the obvious thing\", \"weigh a few approaches\", \"explore alternatives\", or the Explore phase of zirkel-solve.","Anselmoo\u002Fwerkstoff","Anselmoo","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F13209783?v=4","01M799YTB3D7RZ4P3B8CSR2X85","Anselm Hahn","01M799YTBRAM6719D9V0NRGB4R",1,"MIT","https:\u002F\u002Fgithub.com\u002FAnselmoo\u002Fwerkstoff\u002Fblob\u002Fmain\u002Fplugins\u002Fzirkel\u002Fskills\u002Fzirkel-explore-branches\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: zirkel-explore-branches\ndescription: >-\n  Explores multiple genuinely-distinct approaches to a scoped problem in parallel,\n  scores each on Feasibility\u002FImpact\u002FRisk, and selects a winner — so the work does\n  not anchor on the first idea. Use when a task has more than one viable approach\n  and picking the first that comes to mind would be a mistake: \"what are our\n  options here\", \"don't just do the obvious thing\", \"weigh a few approaches\",\n  \"explore alternatives\", or the Explore phase of zirkel-solve.\n---\n\n# zirkel-explore-branches\n\nGenerate branches **independently and in parallel**, score each **in isolation**,\nthen select by the fixed rule. Parallel independence is what prevents anchoring —\nit is structural, not a suggestion.\n\nThe guard is `python3 -B \"${CLAUDE_PLUGIN_ROOT}\u002Fscripts\u002Fzirkel.py\"`, typed in full in each command below.\n\n## 0. Build or reuse the shared research snapshot\n\nBefore dispatching any `branch-proposer`, build (or reuse) zirkel's symbol-index\nsnapshot once, so every parallel proposer can query the same cached index instead\nof each re-scanning the codebase independently:\n\n```\npython3 -B \"${CLAUDE_PLUGIN_ROOT}\u002Fscripts\u002Fbuild_symbol_index.py\" --repo-path . --plugin-name zirkel\n```\n\nThis is a no-op if `analysis\u002Fzirkel\u002Fcurrent.json`'s `source_fingerprint` already\nmatches the repo (see `references\u002Fparallel-safe-research-protocol.md`). Skip this\nstep for a repository well under roughly 50 tracked files — this is a judgment call\nthe script itself does not make (see \"Scale guidance\" in references\u002Fparallel-safe-research-protocol.md).\n\n## Preferred path: the workflow\nWhen the Workflow tool is available, run\n`${CLAUDE_PLUGIN_ROOT}\u002Fworkflows\u002Fexplore-branches.js`\n(args `{ problem, requestedBranches?, maxBranchCount? }`). It dispatches one\n`branch-proposer` agent per angle in parallel (propose), scores each in a separate\ncomparison-free dispatch, and selects the winner — all bounds enforced in code.\n\n**Workflow scripts have no filesystem access.** The workflow returns\n`{ branches, scores, selected, rationale }` in memory and stops there — it\ncannot write `state.json` and cannot render a report. **Run\nthe Persist and Render steps below** after it returns; do not skip them\nbecause \"the workflow already did the work\" silently drops both the\npersisted run and its report, the same \"ends without a trace\" failure the\nManual path would have if its own step 5 were skipped.\n\n## Manual path\n\n### 1. Decide the branch count (guarded)\n`echo '{\"requested\":\u003Cn or null>,\"max_branch_count\":\u003Cfrom config or null>}' | python3 -B \"${CLAUDE_PLUGIN_ROOT}\u002Fscripts\u002Fzirkel.py\" branch-cap -`\n- **Default is 3 branches.** The cap is **min(6, `max_branch_count`)** where\n  `max_branch_count` comes from `.claude\u002Fzirkel.local.md` if present. The guard\n  returns the effective `cap`; produce exactly that many.\n\n### 2. Propose in parallel\nDispatch one `branch-proposer` agent per angle (conservative\u002Fambitious\u002Fpragmatic\u002F…)\n**in a single message so they run concurrently and never see each other.** Each\ncommits fully to its angle. Do not score here.\n\n### 3. Score each in isolation\nDispatch scoring separately — a proposer never scores its own branch, and no\nscorer compares branches. Each returns Feasibility, Impact, Risk (each **1-10**)\nand its biggest blocker.\n\n### 4. Select (guarded)\n```\necho '{\"branches\":[\n  {\"name\":\"A\",\"feasibility\":7,\"impact\":8,\"risk\":4,\"biggest_blocker\":\"…\"},\n  {\"name\":\"B\",\"feasibility\":6,\"impact\":9,\"risk\":6,\"biggest_blocker\":\"…\"}\n]}' | python3 -B \"${CLAUDE_PLUGIN_ROOT}\u002Fscripts\u002Fzirkel.py\" branch-scores -\n```\nThe guard computes **Total = Feasibility + Impact + Risk (Risk NOT inverted)**,\nselects the **highest total**, and **breaks ties by lower risk**. Use its\n`selected` — do not re-derive the winner by hand.\n\n## Persist (so a later `zirkel-solve` can reuse this, and so Render below has\nsomething to render)\n\nRegardless of which path produced the branches and scores above, generate a\nrun id (`python3 -c \"import uuid; print(uuid.uuid4())\"`), then merge each\nbranch's `description` (from Propose \u002F step 2) back onto its scored object\n(from Score+Select \u002F step 3-4, or the workflow's `scores` array) **by name**\nbefore writing — the guard does not require `description`, but the\nbranch-comparison report below does, and this is the only point where the\ntwo ever meet:\n\n```\necho '{\n  \"run_id\": \"\u003Crun-id>\",\n  \"raw_task\": \"\u003Cthe exact `problem` text you were given>\",\n  \"phase\": \"Explore\",\n  \"explore_ran\": true,\n  \"explore\": {\"branches\": [\n    {\"name\":\"A\",\"description\":\"\u003CA'\\''s Propose description>\",\"feasibility\":7,\"impact\":8,\"risk\":4,\"biggest_blocker\":\"…\"},\n    {\"name\":\"B\",\"description\":\"\u003CB'\\''s Propose description>\",\"feasibility\":6,\"impact\":9,\"risk\":6,\"biggest_blocker\":\"…\"}\n  ]}\n}' | python3 -B \"${CLAUDE_PLUGIN_ROOT}\u002Fscripts\u002Fzirkel.py\" state-write - --output-dir .zirkel --to runs\u002F\u003Crun-id>\u002Fstate.json\n```\n\n`raw_task` here is the `problem` text this invocation actually scored branches for.\nA later `zirkel-solve` run only reuses this if its own Explore step is about to\nscore branches for that **exact same text** (see `zirkel-solve`'s own \"Reuse a\nprior run\" step) — matching is byte-for-byte, never fuzzy.\n\n## Render the branch-comparison report\n\nEvery run that reaches Persist also gets a report — not optional, not gated on\nbeing asked for one:\n\n```\npython3 -B \"${CLAUDE_PLUGIN_ROOT}\u002Fscripts\u002Fbuild_branch_comparison_html.py\" . \\\n    --run-id \u003Crun-id> \\\n    --template \"${CLAUDE_PLUGIN_ROOT}\u002Fassets\u002Fbranch-comparison-viewer.html\" \\\n    --d3 \"${CLAUDE_PLUGIN_ROOT}\u002Fassets\u002Finline-d3.html\" \\\n    --tokens \"${CLAUDE_PLUGIN_ROOT}\u002Fassets\u002Ftokens.css\"\n```\n\nThis re-validates `runs\u002F\u003Crun-id>\u002Fstate.json` with the same guard `state-write`\nused to accept it, recomputes Total and the winner from the 3 raw axes (never\ntrusts a stored field — none is persisted), and writes\n`.zirkel\u002Fruns\u002F\u003Crun-id>\u002Fbranch-comparison.html`. Mention this path when\npresenting the selected branch to the user.\n\n## Output\n- branches (name + description)\n- scores table: branch, Feasibility, Impact, Risk, Total, biggest blocker\n- selected branch name + rationale\n- path to the rendered branch-comparison report\n","# zirkel-explore-branches\n\nGenerate branches **independently and in parallel**, score each **in isolation**,\nthen select by the fixed rule. Parallel independence is what prevents anchoring —\nit is structural, not a suggestion.\n\nThe guard is `python3 -B \"${CLAUDE_PLUGIN_ROOT}\u002Fscripts\u002Fzirkel.py\"`, typed in full in each command below.\n\n## 0. Build or reuse the shared research snapshot\n\nBefore dispatching any `branch-proposer`, build (or reuse) zirkel's symbol-index\nsnapshot once, so every parallel proposer can query the same cached index instead\nof each re-scanning the codebase independently:\n\n```\npython3 -B \"${CLAUDE_PLUGIN_ROOT}\u002Fscripts\u002Fbuild_symbol_index.py\" --repo-path . --plugin-name zirkel\n```\n\nThis is a no-op if `analysis\u002Fzirkel\u002Fcurrent.json`'s `source_fingerprint` already\nmatches the repo (see `references\u002Fparallel-safe-research-protocol.md`). Skip this\nstep for a repository well under roughly 50 tracked files — this is a judgment call\nthe script itself does not make (see \"Scale guidance\" in references\u002Fparallel-safe-research-protocol.md).\n\n## Preferred path: the workflow\nWhen the Workflow tool is available, run\n`${CLAUDE_PLUGIN_ROOT}\u002Fworkflows\u002Fexplore-branches.js`\n(args `{ problem, requestedBranches?, maxBranchCount? }`). It dispatches one\n`branch-proposer` agent per angle in parallel (propose), scores each in a separate\ncomparison-free dispatch, and selects the winner — all bounds enforced in code.\n\n**Workflow scripts have no filesystem access.** The workflow returns\n`{ branches, scores, selected, rationale }` in memory and stops there — it\ncannot write `state.json` and cannot render a report. **Run\nthe Persist and Render steps below** after it returns; do not skip them\nbecause \"the workflow already did the work\" silently drops both the\npersisted run and its report, the same \"ends without a trace\" failure the\nManual path would have if its own step 5 were skipped.\n\n## Manual path\n\n### 1. Decide the branch count (guarded)\n`echo '{\"requested\":\u003Cn or null>,\"max_branch_count\":\u003Cfrom config or null>}' | python3 -B \"${CLAUDE_PLUGIN_ROOT}\u002Fscripts\u002Fzirkel.py\" branch-cap -`\n- **Default is 3 branches.** The cap is **min(6, `max_branch_count`)** where\n  `max_branch_count` comes from `.claude\u002Fzirkel.local.md` if present. The guard\n  returns the effective `cap`; produce exactly that many.\n\n### 2. Propose in parallel\nDispatch one `branch-proposer` agent per angle (conservative\u002Fambitious\u002Fpragmatic\u002F…)\n**in a single message so they run concurrently and never see each other.** Each\ncommits fully to its angle. Do not score here.\n\n### 3. Score each in isolation\nDispatch scoring separately — a proposer never scores its own branch, and no\nscorer compares branches. Each returns Feasibility, Impact, Risk (each **1-10**)\nand its biggest blocker.\n\n### 4. Select (guarded)\n```\necho '{\"branches\":[\n  {\"name\":\"A\",\"feasibility\":7,\"impact\":8,\"risk\":4,\"biggest_blocker\":\"…\"},\n  {\"name\":\"B\",\"feasibility\":6,\"impact\":9,\"risk\":6,\"biggest_blocker\":\"…\"}\n]}' | python3 -B \"${CLAUDE_PLUGIN_ROOT}\u002Fscripts\u002Fzirkel.py\" branch-scores -\n```\nThe guard computes **Total = Feasibility + Impact + Risk (Risk NOT inverted)**,\nselects the **highest total**, and **breaks ties by lower risk**. Use its\n`selected` — do not re-derive the winner by hand.\n\n## Persist (so a later `zirkel-solve` can reuse this, and so Render below has\nsomething to render)\n\nRegardless of which path produced the branches and scores above, generate a\nrun id (`python3 -c \"import uuid; print(uuid.uuid4())\"`), then merge each\nbranch's `description` (from Propose \u002F step 2) back onto its scored object\n(from Score+Select \u002F step 3-4, or the workflow's `scores` array) **by name**\nbefore writing — the guard does not require `description`, but the\nbranch-comparison report below does, and this is the only point where the\ntwo ever meet:\n\n```\necho '{\n  \"run_id\": \"\u003Crun-id>\",\n  \"raw_task\": \"\u003Cthe exact `problem` text you were given>\",\n  \"phase\": \"Explore\",\n  \"explore_ran\": true,\n  \"explore\": {\"branches\": [\n    {\"name\":\"A\",\"description\":\"\u003CA'\\''s Propose description>\",\"feasibility\":7,\"impact\":8,\"risk\":4,\"biggest_blocker\":\"…\"},\n    {\"name\":\"B\",\"description\":\"\u003CB'\\''s Propose description>\",\"feasibility\":6,\"impact\":9,\"risk\":6,\"biggest_blocker\":\"…\"}\n  ]}\n}' | python3 -B \"${CLAUDE_PLUGIN_ROOT}\u002Fscripts\u002Fzirkel.py\" state-write - --output-dir .zirkel --to runs\u002F\u003Crun-id>\u002Fstate.json\n```\n\n`raw_task` here is the `problem` text this invocation actually scored branches for.\nA later `zirkel-solve` run only reuses this if its own Explore step is about to\nscore branches for that **exact same text** (see `zirkel-solve`'s own \"Reuse a\nprior run\" step) — matching is byte-for-byte, never fuzzy.\n\n## Render the branch-comparison report\n\nEvery run that reaches Persist also gets a report — not optional, not gated on\nbeing asked for one:\n\n```\npython3 -B \"${CLAUDE_PLUGIN_ROOT}\u002Fscripts\u002Fbuild_branch_comparison_html.py\" . \\\n    --run-id \u003Crun-id> \\\n    --template \"${CLAUDE_PLUGIN_ROOT}\u002Fassets\u002Fbranch-comparison-viewer.html\" \\\n    --d3 \"${CLAUDE_PLUGIN_ROOT}\u002Fassets\u002Finline-d3.html\" \\\n    --tokens \"${CLAUDE_PLUGIN_ROOT}\u002Fassets\u002Ftokens.css\"\n```\n\nThis re-validates `runs\u002F\u003Crun-id>\u002Fstate.json` with the same guard `state-write`\nused to accept it, recomputes Total and the winner from the 3 raw axes (never\ntrusts a stored field — none is persisted), and writes\n`.zirkel\u002Fruns\u002F\u003Crun-id>\u002Fbranch-comparison.html`. Mention this path when\npresenting the selected branch to the user.\n\n## Output\n- branches (name + description)\n- scores table: branch, Feasibility, Impact, Risk, Total, biggest blocker\n- selected branch name + rationale\n- path to the rendered branch-comparison report\n",{"name":5,"description":7},"main","6158c1e4e112c06af2c7ac01b13b2ed7214c0fcf","plugins\u002Fzirkel\u002Fskills\u002Fzirkel-explore-branches",0,"https:\u002F\u002Fgithub.com\u002FAnselmoo\u002Fwerkstoff","https:\u002F\u002Fanselmoo.github.io\u002Fwerkstoff\u002F","https:\u002F\u002Fgithub.com\u002FAnselmoo","Likes to Develop","2026-11-02 04:20:59",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":45},[35,38,41],{"tool":36,"command":37},"npx","npx skills add Anselmoo\u002Fwerkstoff --skill zirkel-explore-branches",{"tool":39,"command":40},"bunx","bunx skills add Anselmoo\u002Fwerkstoff --skill zirkel-explore-branches",{"tool":42,"command":43},"pnpm","pnpm dlx skills add Anselmoo\u002Fwerkstoff --skill zirkel-explore-branches","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FAnselmoo\u002Fwerkstoff\u002Fblob\u002Fmain\u002Fplugins\u002Fzirkel\u002Fskills\u002Fzirkel-explore-branches\u002FSKILL.md",true,[],[48,55,62,67,72,77],{"skillNo":49,"name":50,"nameZh":6,"description":51,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":52,"categories":53,"keywords":54,"publishedAt":19},"01M79AJTRBNBX03WWKYJ8ENM45","toy-check","A skill that checks things in the repository and reports what it finds.","https:\u002F\u002Fgithub.com\u002FAnselmoo\u002Fwerkstoff\u002Fblob\u002Fmain\u002Ftest\u002Fworkflows\u002Ffixtures\u002Ftoy-plugin\u002Fskills\u002Ftoy-check\u002FSKILL.md",[],[],{"skillNo":56,"name":57,"nameZh":6,"description":58,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":59,"categories":60,"keywords":61,"publishedAt":19},"01M79AJS292F3HTDRXWZVRSN5B","demo-check","Reports which generated files changed since the last generation run. Use when the user asks what drifted. Read-only; never edits a file.","https:\u002F\u002Fgithub.com\u002FAnselmoo\u002Fwerkstoff\u002Fblob\u002Fmain\u002Ftest\u002Fplugins\u002Ffixtures\u002Fnacharbeit-uncompleted-run\u002Fplugins\u002Fdemo\u002Fskills\u002Fdemo-check\u002FSKILL.md",[],[],{"skillNo":63,"name":57,"nameZh":6,"description":58,"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},"01M79AJP4TFQXKT1AAK1WC1JYK","https:\u002F\u002Fgithub.com\u002FAnselmoo\u002Fwerkstoff\u002Fblob\u002Fmain\u002Ftest\u002Fplugins\u002Ffixtures\u002Fnacharbeit-stale-lock\u002Fplugins\u002Fdemo\u002Fskills\u002Fdemo-check\u002FSKILL.md",[],[],{"skillNo":68,"name":57,"nameZh":6,"description":58,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":69,"categories":70,"keywords":71,"publishedAt":19},"01M79AJM6Y50WP3G2BR4H0M60Z","https:\u002F\u002Fgithub.com\u002FAnselmoo\u002Fwerkstoff\u002Fblob\u002Fmain\u002Ftest\u002Fplugins\u002Ffixtures\u002Fnacharbeit-preflight-mixed\u002Fplugins\u002Fdemo\u002Fskills\u002Fdemo-check\u002FSKILL.md",[],[],{"skillNo":73,"name":57,"nameZh":6,"description":58,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":74,"categories":75,"keywords":76,"publishedAt":19},"01M79AJGR6DMT8KCCK071V78XM","https:\u002F\u002Fgithub.com\u002FAnselmoo\u002Fwerkstoff\u002Fblob\u002Fmain\u002Ftest\u002Fplugins\u002Ffixtures\u002Fnacharbeit-hooks-shape\u002Fplugins\u002Fdemo\u002Fskills\u002Fdemo-check\u002FSKILL.md",[],[],{"skillNo":78,"name":79,"nameZh":6,"description":80,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":81,"categories":82,"keywords":83,"publishedAt":19},"01M79AJBJ4CHC4PZVWKERWF3HF","zirkel-verify-assumptions","Checks exactly ONE named assumption against real evidence in at most 3 steps and reports one outcome: confidence raised (>=90), reading changed, or still unresolved. Use to resolve a single load-bearing uncertainty — typically a blocking entry surfaced by zirkel-clarify-scope: \"verify this one assumption\", \"is it actually true that X\", \"check this before we rely on it\". For multiple uncertainties, invoke once per uncertainty — the step budget is never shared.","https:\u002F\u002Fgithub.com\u002FAnselmoo\u002Fwerkstoff\u002Fblob\u002Fmain\u002Fplugins\u002Fzirkel\u002Fskills\u002Fzirkel-verify-assumptions\u002FSKILL.md",[],[]]