[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M75BEVSWXHDYHE0WZ1V77F31":3,"skill-same:01M75BEVSWXHDYHE0WZ1V77F31":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":25,"commitSha":26,"skillDirectory":27,"forksCount":28,"repositoryHtmlUrl":29,"repositoryHomepageUrl":6,"ownerHtmlUrl":30,"ownerBlogUrl":31,"ownerDescription":6,"updatedAt":32,"install":33,"resources":46},"01M75BEVSWXHDYHE0WZ1V77F31","guide-decision",null,"Use when the user asks for personal, non-technical advice with no single correct answer - personal finance, veterinary or pet care choices, management or interpersonal decisions, legal-adjacent personal questions, or other life decisions. Also covers method and delivery judgment calls (estimation approach, planning cadence, process tradeoffs). Not for technical\u002Farchitecture work, or questions with one objectively correct answer. Personal purchases of software\u002Fsubscriptions route to `guide-purchase` instead.","NormandErwan\u002FClaude","NormandErwan","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F7470865?v=4","01M75BENGB5CYYFMHFPN2P5VDN","Erwan Normand","01M75BENH549BJRCTY1483T64R",1,"BSD-3-Clause","https:\u002F\u002Fgithub.com\u002FNormandErwan\u002FClaude\u002Fblob\u002Fmain\u002Fskills\u002Fguide-decision\u002FSKILL.md",[],[],"2026-11-12 13:55:43",false,"---\nname: guide-decision\ndescription: Use when the user asks for personal, non-technical advice with no single correct answer - personal finance, veterinary or pet care choices, management or interpersonal decisions, legal-adjacent personal questions, or other life decisions. Also covers method and delivery judgment calls (estimation approach, planning cadence, process tradeoffs). Not for technical\u002Farchitecture work, or questions with one objectively correct answer. Personal purchases of software\u002Fsubscriptions route to `guide-purchase` instead.\nversion: 1.1.0\n---\n\n# Guide Decision\n\n## Overview\n\nAn unframed question has no right answer, and a confident first answer is a\ndraft, not a verdict. Framing before the answer, then a deliberate\nself-review, catches the assumptions and alternatives a single pass misses.\n\n**Core principle:** Never answer a question you have not finished framing -\nthen never ship the first plausible answer either.\n\n## When to Use\n\n- The user asks \"what should I do about X\" for a personal, non-technical matter\n- The user asks for an opinion or recommendation where reasonable experts\n  could disagree\n- The question has real tradeoffs (cost vs. time, risk vs. comfort, short vs.\n  long term)\n\n**Not for:**\n- **Technical\u002Farchitecture** work - how to build, structure, or fix a system.\n  Method and delivery calls (how to estimate, how to plan, which process to\n  run) stay in scope, as does a personal software or subscription purchase,\n  via `guide-purchase`\n- Questions with one factually correct answer (use `verify-sources` instead)\n- Buying decisions (use `guide-purchase` - it reuses the loop below)\n\n## Core Pattern\n\n1. **Frame the need** - a `grilling` gate, run before answering: every\n   ambiguity, unknown, or assumption whose answer would change the\n   recommendation becomes a question, batched by independent branch with a\n   recommended answer each. **Concrete is not specified** - \"how do I service\n   my Peugeot 207\" is a perfectly clear ask that still needs year, engine and\n   trim before any answer, or any verification behind it, holds. Ask for the\n   identifier or the constraint - model reference, existing setup,\n   jurisdiction, horizon - never a photo or a full inventory. Skip only what\n   genuinely cannot move the answer, and don't re-ask what a same-domain\n   followup already established this session.\n2. **Answer as a named domain expert** - state the persona explicitly (\"As a\n   [domain] expert...\") and give a direct, complete first-pass answer. Define\n   every domain term on first use, and carry one concrete worked example\n   through the answer: a term used without definition, or a formula without\n   numbers, is not an expert answer. On a same-domain followup in the same\n   session, don't reintroduce the persona - answer directly using the context\n   already established.\n3. **Pause before finalizing** - take the time to reflect. Don't finalize the\n   first plausible answer; check it actually addresses what was asked.\n4. **Self-critique the draft** - list the assumptions made, and surface 1-3\n   credible alternatives or considerations the first pass missed. Scale the\n   depth to the stakes: a quick personal question needs 1-2 lines, not a\n   multi-section audit; a same-domain followup needs this only if it adds a\n   new tradeoff. When the stakes involve legal exposure, health, or losing\n   something hard to reverse (housing, a job), also flag any specific\n   fact\u002Ffigure\u002Fthreshold stated with confidence in step 2 that isn't\n   verified - point to `verify-sources` or a relevant professional rather\n   than asserting it.\n5. **Revise** - rewrite the answer incorporating the critique. Merge it into\n   a stronger answer; don't tack the critique on as caveats. If the reasoning\n   surfaced a variable you guessed that step 1 could not have known to ask\n   about, ask it now and mark the recommendation provisional on it - not a\n   second round of framing.\n6. **Close with what would change this** - one line naming the condition(s)\n   that would flip the recommendation, so it stays revisable as facts change.\n7. **Consolidate** - when the frame or the recommendation has materially\n   changed across refinement rounds, re-issue the full answer as one document\n   instead of another increment. Deltas accumulate into an unusable answer.\n\n**If the user asks to skip the ceremony** (\"just a direct answer\", \"no\nanalysis\") - steps 3-7 still happen, but stay internal: reply with only the\nrevised answer, dropping the visible critique and closing line. Step 1 is not\ncovered by this: an unasked question that changes the answer is not ceremony.\nDon't skip the critique itself, only its visible form.\n\n## Quick Reference\n\n| Step | Purpose | Skip when |\n|---|---|---|\n| Frame (`grilling`) | Close every ambiguity\u002Funknown\u002Fassumption that moves the answer | Nothing left that moves the answer, or already established this session |\n| Expert answer | Anchor a direct first pass; terms defined, example worked | Persona intro skipped on a same-domain followup (the answer itself never is) |\n| Reflect | Interrupt rushing to the first answer | Never |\n| Self-critique | Surface assumptions + alternatives; verify facts if high-stakes | Never* |\n| Revise | Merge critique into final answer; ask what only the reasoning revealed | Never |\n| What would change it | Keep the recommendation revisable | Never* |\n| Consolidate | Re-issue in full once the frame or recommendation materially moved | The increment still reads on its own |\n\n*Visible form only - stays internal if the user asked for a direct answer.\n\n## Example\n\n> User: \"Should I put my dog on a raw diet?\"\n>\n> Frame first - three answers change the recommendation, so they get asked\n> before it is written, not assumed:\n> - Age and health status? *(Recommended answer: healthy adult. Puppies,\n>   seniors and immunocompromised dogs flip the risk calculus.)*\n> - Has a vet already recommended raw for a specific condition?\n>   *(Recommended: no - if yes, their guidance supersedes this.)*\n> - Anyone immunocompromised, pregnant, or under 5 in the household?\n>   *(Recommended: no - it drives the handling risk, not the dog's.)*\n>\n> User: healthy 4-year-old, no vet recommendation, two adults at home.\n>\n> As a veterinary nutrition expert: raw diets carry salmonella\u002Flisteria risk\n> for the pet and household, and are rarely nutritionally complete without\n> careful formulation - I'd recommend a commercial diet formulated to AAFCO\n> standards instead, with real meat high on the ingredient list.\n>\n> *Alternative not yet covered: a vet-supervised, commercially-balanced raw\n> diet avoids most of the nutritional-completeness risk, if raw is important\n> to the owner.*\n>\n> Revised: for a healthy dog, a quality AAFCO-formulated commercial diet is\n> the lower-risk default. If raw still matters to you, do it only with a\n> vet-formulated recipe (not homemade improvisation) to avoid the deficiency\n> risk that makes raw diets risky in the first place.\n>\n> *This would change if a vet has already diagnosed a condition raw feeding\n> is meant to address - follow their specific guidance instead.*\n\n## Common Mistakes\n\n- Treating a concrete ask as a specified one - skipping the frame because the\n  question reads clearly, when the answer still pivots on an unstated model,\n  year, variant, or existing setup\n- Answering about a specific instance from a generic mental model when\n  variants differ - describing a mechanism, or costing a setup, the user's\n  actual model may not have\n- Stacking delta after delta across refinement rounds without ever re-issuing\n  a consolidated answer\n- Self-critiquing but not actually revising - the critique becomes a hedge\n  appended to the original answer instead of changing it\n- Naming a persona without answering as that persona - \"As a vet, it depends\"\n  isn't an expert answer\n- Producing alternatives that are cosmetic variations of the same answer,\n  not genuinely different considerations\n- Writing a multi-section visible audit trail for a low-stakes question -\n  match the critique's length to how much the decision matters\n- Treating \"just a direct answer\" as license to skip the internal critique\n  itself, not just its visible form\n- Re-framing or reintroducing the persona on a same-domain followup already\n  established earlier in the session\n- Stating a specific legal threshold, dosage, or figure with confidence on a\n  high-stakes question without flagging it for verification\n","# Guide Decision\n\n## Overview\n\nAn unframed question has no right answer, and a confident first answer is a\ndraft, not a verdict. Framing before the answer, then a deliberate\nself-review, catches the assumptions and alternatives a single pass misses.\n\n**Core principle:** Never answer a question you have not finished framing -\nthen never ship the first plausible answer either.\n\n## When to Use\n\n- The user asks \"what should I do about X\" for a personal, non-technical matter\n- The user asks for an opinion or recommendation where reasonable experts\n  could disagree\n- The question has real tradeoffs (cost vs. time, risk vs. comfort, short vs.\n  long term)\n\n**Not for:**\n- **Technical\u002Farchitecture** work - how to build, structure, or fix a system.\n  Method and delivery calls (how to estimate, how to plan, which process to\n  run) stay in scope, as does a personal software or subscription purchase,\n  via `guide-purchase`\n- Questions with one factually correct answer (use `verify-sources` instead)\n- Buying decisions (use `guide-purchase` - it reuses the loop below)\n\n## Core Pattern\n\n1. **Frame the need** - a `grilling` gate, run before answering: every\n   ambiguity, unknown, or assumption whose answer would change the\n   recommendation becomes a question, batched by independent branch with a\n   recommended answer each. **Concrete is not specified** - \"how do I service\n   my Peugeot 207\" is a perfectly clear ask that still needs year, engine and\n   trim before any answer, or any verification behind it, holds. Ask for the\n   identifier or the constraint - model reference, existing setup,\n   jurisdiction, horizon - never a photo or a full inventory. Skip only what\n   genuinely cannot move the answer, and don't re-ask what a same-domain\n   followup already established this session.\n2. **Answer as a named domain expert** - state the persona explicitly (\"As a\n   [domain] expert...\") and give a direct, complete first-pass answer. Define\n   every domain term on first use, and carry one concrete worked example\n   through the answer: a term used without definition, or a formula without\n   numbers, is not an expert answer. On a same-domain followup in the same\n   session, don't reintroduce the persona - answer directly using the context\n   already established.\n3. **Pause before finalizing** - take the time to reflect. Don't finalize the\n   first plausible answer; check it actually addresses what was asked.\n4. **Self-critique the draft** - list the assumptions made, and surface 1-3\n   credible alternatives or considerations the first pass missed. Scale the\n   depth to the stakes: a quick personal question needs 1-2 lines, not a\n   multi-section audit; a same-domain followup needs this only if it adds a\n   new tradeoff. When the stakes involve legal exposure, health, or losing\n   something hard to reverse (housing, a job), also flag any specific\n   fact\u002Ffigure\u002Fthreshold stated with confidence in step 2 that isn't\n   verified - point to `verify-sources` or a relevant professional rather\n   than asserting it.\n5. **Revise** - rewrite the answer incorporating the critique. Merge it into\n   a stronger answer; don't tack the critique on as caveats. If the reasoning\n   surfaced a variable you guessed that step 1 could not have known to ask\n   about, ask it now and mark the recommendation provisional on it - not a\n   second round of framing.\n6. **Close with what would change this** - one line naming the condition(s)\n   that would flip the recommendation, so it stays revisable as facts change.\n7. **Consolidate** - when the frame or the recommendation has materially\n   changed across refinement rounds, re-issue the full answer as one document\n   instead of another increment. Deltas accumulate into an unusable answer.\n\n**If the user asks to skip the ceremony** (\"just a direct answer\", \"no\nanalysis\") - steps 3-7 still happen, but stay internal: reply with only the\nrevised answer, dropping the visible critique and closing line. Step 1 is not\ncovered by this: an unasked question that changes the answer is not ceremony.\nDon't skip the critique itself, only its visible form.\n\n## Quick Reference\n\n| Step | Purpose | Skip when |\n|---|---|---|\n| Frame (`grilling`) | Close every ambiguity\u002Funknown\u002Fassumption that moves the answer | Nothing left that moves the answer, or already established this session |\n| Expert answer | Anchor a direct first pass; terms defined, example worked | Persona intro skipped on a same-domain followup (the answer itself never is) |\n| Reflect | Interrupt rushing to the first answer | Never |\n| Self-critique | Surface assumptions + alternatives; verify facts if high-stakes | Never* |\n| Revise | Merge critique into final answer; ask what only the reasoning revealed | Never |\n| What would change it | Keep the recommendation revisable | Never* |\n| Consolidate | Re-issue in full once the frame or recommendation materially moved | The increment still reads on its own |\n\n*Visible form only - stays internal if the user asked for a direct answer.\n\n## Example\n\n> User: \"Should I put my dog on a raw diet?\"\n>\n> Frame first - three answers change the recommendation, so they get asked\n> before it is written, not assumed:\n> - Age and health status? *(Recommended answer: healthy adult. Puppies,\n>   seniors and immunocompromised dogs flip the risk calculus.)*\n> - Has a vet already recommended raw for a specific condition?\n>   *(Recommended: no - if yes, their guidance supersedes this.)*\n> - Anyone immunocompromised, pregnant, or under 5 in the household?\n>   *(Recommended: no - it drives the handling risk, not the dog's.)*\n>\n> User: healthy 4-year-old, no vet recommendation, two adults at home.\n>\n> As a veterinary nutrition expert: raw diets carry salmonella\u002Flisteria risk\n> for the pet and household, and are rarely nutritionally complete without\n> careful formulation - I'd recommend a commercial diet formulated to AAFCO\n> standards instead, with real meat high on the ingredient list.\n>\n> *Alternative not yet covered: a vet-supervised, commercially-balanced raw\n> diet avoids most of the nutritional-completeness risk, if raw is important\n> to the owner.*\n>\n> Revised: for a healthy dog, a quality AAFCO-formulated commercial diet is\n> the lower-risk default. If raw still matters to you, do it only with a\n> vet-formulated recipe (not homemade improvisation) to avoid the deficiency\n> risk that makes raw diets risky in the first place.\n>\n> *This would change if a vet has already diagnosed a condition raw feeding\n> is meant to address - follow their specific guidance instead.*\n\n## Common Mistakes\n\n- Treating a concrete ask as a specified one - skipping the frame because the\n  question reads clearly, when the answer still pivots on an unstated model,\n  year, variant, or existing setup\n- Answering about a specific instance from a generic mental model when\n  variants differ - describing a mechanism, or costing a setup, the user's\n  actual model may not have\n- Stacking delta after delta across refinement rounds without ever re-issuing\n  a consolidated answer\n- Self-critiquing but not actually revising - the critique becomes a hedge\n  appended to the original answer instead of changing it\n- Naming a persona without answering as that persona - \"As a vet, it depends\"\n  isn't an expert answer\n- Producing alternatives that are cosmetic variations of the same answer,\n  not genuinely different considerations\n- Writing a multi-section visible audit trail for a low-stakes question -\n  match the critique's length to how much the decision matters\n- Treating \"just a direct answer\" as license to skip the internal critique\n  itself, not just its visible form\n- Re-framing or reintroducing the persona on a same-domain followup already\n  established earlier in the session\n- Stating a specific legal threshold, dosage, or figure with confidence on a\n  high-stakes question without flagging it for verification\n",{"name":5,"version":24,"description":7},"1.1.0","main","336ef9f9a6dd71a2cc287082873fa3c8597dcf75","skills\u002Fguide-decision",0,"https:\u002F\u002Fgithub.com\u002FNormandErwan\u002FClaude","https:\u002F\u002Fgithub.com\u002FNormandErwan","https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Fnormanderwan\u002F","2026-11-11 04:37:11",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":45},[35,38,41],{"tool":36,"command":37},"npx","npx skills add NormandErwan\u002FClaude --skill guide-decision",{"tool":39,"command":40},"bunx","bunx skills add NormandErwan\u002FClaude --skill guide-decision",{"tool":42,"command":43},"pnpm","pnpm dlx skills add NormandErwan\u002FClaude --skill guide-decision","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FNormandErwan\u002FClaude\u002Fblob\u002Fmain\u002Fskills\u002Fguide-decision\u002FSKILL.md",true,[],[48,55,62,69,76,83],{"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},"01M75BG0TBQH53V52ZF840MGD2","write-guide","Use when writing a guide someone else will follow - a method guide, a playbook, a how-we-do-it document for a team or a client - or when reworking, auditing or extending one that already exists. Also when an outline is handed to you to fill in, when a guide's rules need evidence labels, or when a second guide joins the first and the two start contradicting each other.","https:\u002F\u002Fgithub.com\u002FNormandErwan\u002FClaude\u002Fblob\u002Fmain\u002Fskills\u002Fwrite-guide\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},"01M75BFX09MYSAKVZCM1713YAE","write-french","Use when the output is French - a chat reply in French, or a French markdown deliverable such as a guide, a report or a synthesis - including when the user writes in French and expects French back. Prevents French that reads as translated English: word-for-word calques, sentences compressed until the meaning is gone, jargon used before it is defined, invented turns of phrase.","https:\u002F\u002Fgithub.com\u002FNormandErwan\u002FClaude\u002Fblob\u002Fmain\u002Fskills\u002Fwrite-french\u002FSKILL.md",[],[],{"skillNo":63,"name":64,"nameZh":6,"description":65,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":66,"categories":67,"keywords":68,"publishedAt":19},"01M75BFQQ0EPV92467E0C40CEX","verify-sources","Use when about to state a factual, technical, historical, or procedural claim in ANY domain (software, DIY\u002Fcraft, health, legal, finance...) - a version number, default value, statistic, API behavior, a legal deadline, a health claim, a manufacturer spec, \"X always\u002Fnever does Y\", \"the docs say\", \"best practice is\" - that you have not verified against a source this turn, or when tempted to answer a factual question from memory under time pressure","https:\u002F\u002Fgithub.com\u002FNormandErwan\u002FClaude\u002Fblob\u002Fmain\u002Fskills\u002Fverify-sources\u002FSKILL.md",[],[],{"skillNo":70,"name":71,"nameZh":6,"description":72,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":73,"categories":74,"keywords":75,"publishedAt":19},"01M75BFJW35J7NT1MGEZ0P477T","token-file-ops","Use when reading, editing, creating, or manipulating files; or running commands with potentially large output. Provides generic bash-first patterns for any file type.","https:\u002F\u002Fgithub.com\u002FNormandErwan\u002FClaude\u002Fblob\u002Fmain\u002Fskills\u002Ftoken-file-ops\u002FSKILL.md",[],[],{"skillNo":77,"name":78,"nameZh":6,"description":79,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":80,"categories":81,"keywords":82,"publishedAt":19},"01M75BFEFWPR9ZXEWACJJBQBRS","token-codebase-exploration","Use when exploring an unfamiliar codebase, understanding how a specific feature or pattern works, or learning how components interact. Provides the exploration workflow and strategy.","https:\u002F\u002Fgithub.com\u002FNormandErwan\u002FClaude\u002Fblob\u002Fmain\u002Fskills\u002Ftoken-codebase-exploration\u002FSKILL.md",[],[],{"skillNo":84,"name":85,"nameZh":6,"description":86,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":87,"categories":88,"keywords":89,"publishedAt":19},"01M75BFBQA8JWX7TFPDWTCTFW2","teach","Use when the user sets learning a topic or skill as their own goal - \"teach me X\", \"I want to learn X\", \"help me get good at X\", a course or study plan for themselves. Not for a question one reply answers (\"what is X\", \"explain X\").","https:\u002F\u002Fgithub.com\u002FNormandErwan\u002FClaude\u002Fblob\u002Fmain\u002Fskills\u002Fteach\u002FSKILL.md",[],[]]