[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6VCATJZCA7ACAJKC5ZPCD2E":3,"skill-same:01M6VCATJZCA7ACAJKC5ZPCD2E":63},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":15,"categories":16,"keywords":17,"publishedAt":26,"linkOnly":27,"markdownContent":28,"markdownBody":29,"frontMatter":30,"refName":33,"commitSha":34,"skillDirectory":35,"forksCount":13,"repositoryHtmlUrl":36,"repositoryHomepageUrl":6,"ownerHtmlUrl":37,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":38,"install":39,"resources":52},"01M6VCATJZCA7ACAJKC5ZPCD2E","team-builder",null,"Interactive agent picker for composing and dispatching parallel teams","sakamoto-family-smile\u002Fagent_monorepo","sakamoto-family-smile","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F132812403?v=4","01M6RTAT3H8PMWX5SJ65KP3VYF","01M6RTAV321WB5RAHCWTZMX5CV",0,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Fsakamoto-family-smile\u002Fagent_monorepo\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fecc\u002Fteam-builder\u002FSKILL.md",[],[18,19,20,21,22,23,24,25],"team","builder","team builder","teams","parallel","interactive","community","for","2026-11-10 10:12:30",false,"---\nname: adu-corrections-flow\ndescription: Analyzes ADU permit corrections letters — the first half of the corrections pipeline. Reads the corrections letter, builds a sheet manifest from the plan binder, researches state and city codes, views referenced plan sheets, categorizes each correction item, and generates informed contractor questions. This skill should be used when a contractor receives a city corrections letter for an ADU permit. It coordinates three sub-skills (california-adu for state law, adu-city-research for city rules, adu-targeted-page-viewer for plan sheet navigation) to produce research artifacts and a UI-ready questions JSON. Does NOT generate the final response package — that is handled by adu-corrections-complete after the contractor answers questions. Triggers when a corrections letter PDF\u002FPNG is provided along with the plan binder PDF.\n---\n\n# ADU Corrections Flow\n\n## Overview\n\nAnalyze ADU permit corrections and generate informed contractor questions. This is the first skill in a two-skill pipeline:\n\n1. **`adu-corrections-flow`** (this skill) — reads corrections, researches codes, categorizes items, generates questions\n2. **`adu-corrections-complete`** (second skill) — takes contractor answers + these research artifacts, generates the final response package\n\nThis skill coordinates three sub-skills through a 4-phase workflow and stops after producing `contractor_questions.json`.\n\n**Sub-skills used:**\n\n| Skill | Role | When Used |\n|-------|------|-----------|\n| `california-adu` | State-level building codes (CRC, CBC, CPC, etc.) | Phase 3A — offline, 28 reference files |\n| `adu-city-research` | City municipal code, standard details, IBs | Phase 3B (Mode 1: Discovery) + Phase 3.5 (Mode 2: Extraction) + optional Mode 3 (Browser Fallback) |\n| `adu-targeted-page-viewer` | Sheet manifest + on-demand plan viewing | Phase 2 + Phase 3C — PDF extraction + vision |\n\n**Key principle:** Research happens *before* contractor questions. Questions informed by actual code requirements are specific and answerable in seconds. Vague questions waste the contractor's time.\n\n## Inputs\n\n| Input | Format | Required |\n|-------|--------|----------|\n| Corrections letter | PDF or PNG (1-3 pages) | Yes |\n| Plan binder | PDF (the full construction plan set) | Yes |\n| City name | String (extracted from letter if not provided) | Auto-detected |\n| Project address | String (extracted from letter if not provided) | Auto-detected |\n\n## Outputs\n\nAll outputs are written to the session directory (e.g., `correction-01\u002F`).\n\n| Output | Format | Phase |\n|--------|--------|-------|\n| `corrections_parsed.json` | Structured correction items | Phase 1 |\n| `sheet-manifest.json` | Sheet ID ↔ page number mapping | Phase 2 |\n| `state_law_findings.json` | Per-code-section lookups | Phase 3A |\n| `city_discovery.json` | Key URLs for the city's ADU pages | Phase 3B |\n| `sheet_observations.json` | What's on each referenced plan sheet | Phase 3C |\n| `city_research_findings.json` | Municipal code, standard details, IBs (extracted content) | Phase 3.5 |\n| `corrections_categorized.json` | Items with categories + research context (the main handoff artifact) | Phase 4 |\n| `contractor_questions.json` | UI-ready question form data | Phase 4 |\n\n**This skill stops here.** The `contractor_questions.json` goes to the UI. After the contractor answers, the `adu-corrections-complete` skill takes the session directory + `contractor_answers.json` and generates the final response package (response letter, professional scope, corrections report, sheet annotations).\n\n**Do NOT generate Phase 5 outputs** (response letter, professional scope, etc.). That is the job of `adu-corrections-complete`. Generating them here creates TODO-filled drafts that the second skill doesn't use.\n\n## Workflow\n\n### Phase 1 + Phase 2 (concurrent)\n\nThese two phases run simultaneously — they have no dependencies on each other.\n\n#### Phase 1: Read Corrections Letter\n\nRead the corrections letter visually (1-3 page PNG or PDF). No sub-skill needed — direct vision reading.\n\nExtract each correction item as a structured object. Preserve the exact original wording. Identify all code references (CRC, CBC, ASCE, B&P Code, municipal code, etc.) and any sheet references.\n\nSave as `corrections_parsed.json`. See `references\u002Foutput-schemas.md` for the full schema.\n\n#### Phase 2: Build Sheet Manifest\n\nRun the `adu-targeted-page-viewer` skill workflow:\n\n1. **Check first:** PNGs and title block crops may already be pre-extracted in `project-files\u002Fpages-png\u002F` and `project-files\u002Ftitle-blocks\u002F`. If they exist, skip extraction and go straight to reading the cover sheet.\n2. If PNGs don't exist: Extract PDF pages to PNGs: `scripts\u002Fextract-pages.sh \u003Cbinder.pdf> \u003Coutput-dir>`\n3. Read the cover sheet (page 1) for the sheet index\n4. If page count differs from index count, crop and read title blocks to resolve\n5. Save `sheet-manifest.json`\n\nThis takes ~90 seconds (or ~30 seconds if PNGs are pre-extracted) and produces the sheet-to-page mapping needed for Phase 3C.\n\n### Phase 3 (concurrent — 3 subagents)\n\nAfter Phases 1+2 complete, launch three parallel research subagents. Each is specialized by domain. All receive the parsed corrections from Phase 1.\n\nSee `references\u002Fsubagent-prompts.md` for the full subagent prompts.\n\n#### Subagent 3A: State Law Researcher\n\n- **Skill context:** `california-adu` (28 reference files, all offline)\n- **Input:** All correction items with their code references\n- **Task:** Look up every referenced code section. Deduplicate — if multiple items cite the same section, look it up once and link to all relevant items.\n- **Speed:** Fast — no network, just reading reference files (~60 sec)\n- **Output:** Per-code-section findings with requirements, thresholds, ADU exceptions\n\n#### Subagent 3B: City Discovery\n\n- **Skill context:** `adu-city-research` — **Mode 1 (Discovery) only**\n- **Input:** City name + list of topics extracted from corrections\n- **Task:** Run WebSearch to find the city's key ADU-related URLs: ADU page, municipal code platform, standard detail PDFs, Information Bulletins, submittal requirements. Do NOT fetch page content — just find URLs.\n- **Speed:** Fast — WebSearch only (~30 sec)\n- **Output:** `city_discovery.json` — categorized URL list for extraction\n\n#### Subagent 3C: Sheet Viewer\n\n- **Skill context:** `adu-targeted-page-viewer`\n- **Input:** Sheet manifest from Phase 2 + sheet references from corrections\n- **Task:** Read only the plan sheets referenced by correction items (typically 5-8 out of 15-30 pages). For each, describe what is currently drawn in the area relevant to the correction.\n- **Speed:** Fast — just reading PNGs (~60 sec)\n- **Output:** Per-sheet observations: current state, what appears missing, location on sheet\n\n### Phase 3.5: City Extraction\n\nAfter Phase 3 completes (all three subagents return), launch city content extraction using the URLs discovered by Subagent 3B.\n\n#### Single-Agent Mode (default)\n\nOne subagent runs `adu-city-research` **Mode 2 (Targeted Extraction)** against all discovered URLs.\n\n- **Input:** `city_discovery.json` + correction topics\n- **Task:** WebFetch each discovered URL, extract content relevant to corrections. Prioritize standard detail PDFs and municipal code sections.\n- **Speed:** ~60-90 sec (depends on URL count)\n- **Output:** `city_research_findings.json`\n\n#### Fan-Out Mode (optional, for speed)\n\nSplit the discovered URLs across 2-3 subagents by topic:\n\n- **Agent 1:** Municipal code URLs — local amendments, ADU ordinance, grading chapter\n- **Agent 2:** Standard detail PDFs + Information Bulletin URLs\n- **Agent 3:** ADU page + submittal requirements\n\nEach agent runs `adu-city-research` Mode 2 with its URL subset. Orchestrator merges results into a single `city_research_findings.json`.\n\n**When to use fan-out:** When Discovery returns 6+ URLs across multiple categories. For smaller cities with 2-3 URLs, single-agent is sufficient.\n\n#### Browser Fallback (conditional)\n\nIf Mode 2 extraction has gaps (URLs that returned empty, PDFs that couldn't be read, sections not found), launch one subagent running `adu-city-research` **Mode 3 (Browser Fallback)** with Chrome MCP.\n\n- **Input:** `extraction_gaps` from Mode 2 output\n- **Task:** Navigate the city's website with browser automation to fill specific gaps\n- **Speed:** ~2-3 min\n- **Output:** Gap-filling additions merged into `city_research_findings.json`\n\n**Only run Browser Fallback if there are actionable gaps.** Most cities' information is accessible via WebSearch + WebFetch. Browser Fallback is for the edge cases.\n\n### Phase 4: Merge + Categorize + Generate Questions\n\nSingle agent merges all three research streams and does the intelligence work.\n\n**For each correction item, cross-reference:**\n1. What does the correction letter say? (Phase 1)\n2. What does state law require? (Phase 3A)\n3. Does the city add anything? (Phase 3.5)\n4. What's currently on the plan sheet? (Phase 3C)\n\n**Then categorize:**\n\n| Category | Meaning | Example |\n|----------|---------|---------|\n| `AUTO_FIXABLE` | Resolve by adding notes, marking checklists, updating labels | Missing CalGreen item, governing codes list |\n| `NEEDS_CONTRACTOR_INPUT` | Requires specific facts from the contractor | Sewer line size, finished grade elevations |\n| `NEEDS_PROFESSIONAL` | Requires licensed professional work (designer, engineer, HERS rater) | Structural calcs, fire-rated assembly detail |\n\n**Then generate questions** for `NEEDS_CONTRACTOR_INPUT` items. Each question includes `research_context` explaining why it's being asked and what the code requires. See `references\u002Foutput-schemas.md` for the `contractor_questions.json` schema.\n\n**Output files:** `corrections_categorized.json` + `contractor_questions.json`\n\n**Return `contractor_questions.json` to the UI.** This skill is now complete. Stop here.\n\n**What happens next:** The UI renders the questions. The contractor answers. Then the `adu-corrections-complete` skill takes the session directory + `contractor_answers.json` and generates the response package. That is a separate agent invocation — not a continuation of this one.\n\n## Timing\n\n| Phase | Time | Notes |\n|-------|------|-------|\n| Phase 1 | ~30 sec | Vision reading, 1-3 pages |\n| Phase 2 | ~90 sec | PDF extraction + manifest building |\n| Phase 3A | ~60 sec | Offline reference lookup |\n| Phase 3B | ~30 sec | City URL discovery (WebSearch only) |\n| Phase 3C | ~60 sec | Reading 5-8 PNGs |\n| Phase 3.5 | ~60-90 sec | City content extraction (WebFetch) |\n| Phase 3.5-fallback | ~2-3 min | Browser fallback (only if needed) |\n| Phase 4 | ~2 min | Merge + categorize + questions |\n| **Total (no fallback)** | **~4-5 min** | **Typical case** |\n| **Total (with fallback)** | **~6-8 min** | **Difficult city website** |\n\n## Important Notes\n\n- **This skill stops after Phase 4.** Do NOT generate response letters, professional scopes, or any Phase 5 outputs. That is the job of `adu-corrections-complete`.\n- **Research before questions.** Never generate contractor questions without first doing code research. The research makes the questions specific and actionable.\n- **Write high-quality research artifacts.** The `corrections_categorized.json` is the main handoff to the second skill. Every item must have its research context, code findings, and sheet observations fully documented — because the second skill runs cold with no conversation history.\n- **Sheet references are sacred.** Every sheet reference must come from `sheet-manifest.json`. Never guess.\n- **This tool helps contractors comply, not litigate.** Focus on *how to fix it*, not whether the correction is valid. If the city says fix it, help the contractor fix it.\n- **City research uses two passes.** Phase 3B (Discovery) runs fast via WebSearch in parallel with 3A\u002F3C. Phase 3.5 (Extraction) uses WebFetch against discovered URLs. Browser Fallback only runs if WebFetch has gaps. This two-pass approach cuts city research from ~5 min to ~90 sec for most cities.\n\n## References\n\n| File | Contents |\n|------|----------|\n| `references\u002Foutput-schemas.md` | JSON schemas for all output files — corrections_parsed, contractor_questions, contractor_answers |\n| `references\u002Fsubagent-prompts.md` | Full prompts for Phase 3 subagents (state law, city research, sheet viewer) + Phase 4 merge prompt |\n","# ADU Corrections Flow\n\n## Overview\n\nAnalyze ADU permit corrections and generate informed contractor questions. This is the first skill in a two-skill pipeline:\n\n1. **`adu-corrections-flow`** (this skill) — reads corrections, researches codes, categorizes items, generates questions\n2. **`adu-corrections-complete`** (second skill) — takes contractor answers + these research artifacts, generates the final response package\n\nThis skill coordinates three sub-skills through a 4-phase workflow and stops after producing `contractor_questions.json`.\n\n**Sub-skills used:**\n\n| Skill | Role | When Used |\n|-------|------|-----------|\n| `california-adu` | State-level building codes (CRC, CBC, CPC, etc.) | Phase 3A — offline, 28 reference files |\n| `adu-city-research` | City municipal code, standard details, IBs | Phase 3B (Mode 1: Discovery) + Phase 3.5 (Mode 2: Extraction) + optional Mode 3 (Browser Fallback) |\n| `adu-targeted-page-viewer` | Sheet manifest + on-demand plan viewing | Phase 2 + Phase 3C — PDF extraction + vision |\n\n**Key principle:** Research happens *before* contractor questions. Questions informed by actual code requirements are specific and answerable in seconds. Vague questions waste the contractor's time.\n\n## Inputs\n\n| Input | Format | Required |\n|-------|--------|----------|\n| Corrections letter | PDF or PNG (1-3 pages) | Yes |\n| Plan binder | PDF (the full construction plan set) | Yes |\n| City name | String (extracted from letter if not provided) | Auto-detected |\n| Project address | String (extracted from letter if not provided) | Auto-detected |\n\n## Outputs\n\nAll outputs are written to the session directory (e.g., `correction-01\u002F`).\n\n| Output | Format | Phase |\n|--------|--------|-------|\n| `corrections_parsed.json` | Structured correction items | Phase 1 |\n| `sheet-manifest.json` | Sheet ID ↔ page number mapping | Phase 2 |\n| `state_law_findings.json` | Per-code-section lookups | Phase 3A |\n| `city_discovery.json` | Key URLs for the city's ADU pages | Phase 3B |\n| `sheet_observations.json` | What's on each referenced plan sheet | Phase 3C |\n| `city_research_findings.json` | Municipal code, standard details, IBs (extracted content) | Phase 3.5 |\n| `corrections_categorized.json` | Items with categories + research context (the main handoff artifact) | Phase 4 |\n| `contractor_questions.json` | UI-ready question form data | Phase 4 |\n\n**This skill stops here.** The `contractor_questions.json` goes to the UI. After the contractor answers, the `adu-corrections-complete` skill takes the session directory + `contractor_answers.json` and generates the final response package (response letter, professional scope, corrections report, sheet annotations).\n\n**Do NOT generate Phase 5 outputs** (response letter, professional scope, etc.). That is the job of `adu-corrections-complete`. Generating them here creates TODO-filled drafts that the second skill doesn't use.\n\n## Workflow\n\n### Phase 1 + Phase 2 (concurrent)\n\nThese two phases run simultaneously — they have no dependencies on each other.\n\n#### Phase 1: Read Corrections Letter\n\nRead the corrections letter visually (1-3 page PNG or PDF). No sub-skill needed — direct vision reading.\n\nExtract each correction item as a structured object. Preserve the exact original wording. Identify all code references (CRC, CBC, ASCE, B&P Code, municipal code, etc.) and any sheet references.\n\nSave as `corrections_parsed.json`. See `references\u002Foutput-schemas.md` for the full schema.\n\n#### Phase 2: Build Sheet Manifest\n\nRun the `adu-targeted-page-viewer` skill workflow:\n\n1. **Check first:** PNGs and title block crops may already be pre-extracted in `project-files\u002Fpages-png\u002F` and `project-files\u002Ftitle-blocks\u002F`. If they exist, skip extraction and go straight to reading the cover sheet.\n2. If PNGs don't exist: Extract PDF pages to PNGs: `scripts\u002Fextract-pages.sh \u003Cbinder.pdf> \u003Coutput-dir>`\n3. Read the cover sheet (page 1) for the sheet index\n4. If page count differs from index count, crop and read title blocks to resolve\n5. Save `sheet-manifest.json`\n\nThis takes ~90 seconds (or ~30 seconds if PNGs are pre-extracted) and produces the sheet-to-page mapping needed for Phase 3C.\n\n### Phase 3 (concurrent — 3 subagents)\n\nAfter Phases 1+2 complete, launch three parallel research subagents. Each is specialized by domain. All receive the parsed corrections from Phase 1.\n\nSee `references\u002Fsubagent-prompts.md` for the full subagent prompts.\n\n#### Subagent 3A: State Law Researcher\n\n- **Skill context:** `california-adu` (28 reference files, all offline)\n- **Input:** All correction items with their code references\n- **Task:** Look up every referenced code section. Deduplicate — if multiple items cite the same section, look it up once and link to all relevant items.\n- **Speed:** Fast — no network, just reading reference files (~60 sec)\n- **Output:** Per-code-section findings with requirements, thresholds, ADU exceptions\n\n#### Subagent 3B: City Discovery\n\n- **Skill context:** `adu-city-research` — **Mode 1 (Discovery) only**\n- **Input:** City name + list of topics extracted from corrections\n- **Task:** Run WebSearch to find the city's key ADU-related URLs: ADU page, municipal code platform, standard detail PDFs, Information Bulletins, submittal requirements. Do NOT fetch page content — just find URLs.\n- **Speed:** Fast — WebSearch only (~30 sec)\n- **Output:** `city_discovery.json` — categorized URL list for extraction\n\n#### Subagent 3C: Sheet Viewer\n\n- **Skill context:** `adu-targeted-page-viewer`\n- **Input:** Sheet manifest from Phase 2 + sheet references from corrections\n- **Task:** Read only the plan sheets referenced by correction items (typically 5-8 out of 15-30 pages). For each, describe what is currently drawn in the area relevant to the correction.\n- **Speed:** Fast — just reading PNGs (~60 sec)\n- **Output:** Per-sheet observations: current state, what appears missing, location on sheet\n\n### Phase 3.5: City Extraction\n\nAfter Phase 3 completes (all three subagents return), launch city content extraction using the URLs discovered by Subagent 3B.\n\n#### Single-Agent Mode (default)\n\nOne subagent runs `adu-city-research` **Mode 2 (Targeted Extraction)** against all discovered URLs.\n\n- **Input:** `city_discovery.json` + correction topics\n- **Task:** WebFetch each discovered URL, extract content relevant to corrections. Prioritize standard detail PDFs and municipal code sections.\n- **Speed:** ~60-90 sec (depends on URL count)\n- **Output:** `city_research_findings.json`\n\n#### Fan-Out Mode (optional, for speed)\n\nSplit the discovered URLs across 2-3 subagents by topic:\n\n- **Agent 1:** Municipal code URLs — local amendments, ADU ordinance, grading chapter\n- **Agent 2:** Standard detail PDFs + Information Bulletin URLs\n- **Agent 3:** ADU page + submittal requirements\n\nEach agent runs `adu-city-research` Mode 2 with its URL subset. Orchestrator merges results into a single `city_research_findings.json`.\n\n**When to use fan-out:** When Discovery returns 6+ URLs across multiple categories. For smaller cities with 2-3 URLs, single-agent is sufficient.\n\n#### Browser Fallback (conditional)\n\nIf Mode 2 extraction has gaps (URLs that returned empty, PDFs that couldn't be read, sections not found), launch one subagent running `adu-city-research` **Mode 3 (Browser Fallback)** with Chrome MCP.\n\n- **Input:** `extraction_gaps` from Mode 2 output\n- **Task:** Navigate the city's website with browser automation to fill specific gaps\n- **Speed:** ~2-3 min\n- **Output:** Gap-filling additions merged into `city_research_findings.json`\n\n**Only run Browser Fallback if there are actionable gaps.** Most cities' information is accessible via WebSearch + WebFetch. Browser Fallback is for the edge cases.\n\n### Phase 4: Merge + Categorize + Generate Questions\n\nSingle agent merges all three research streams and does the intelligence work.\n\n**For each correction item, cross-reference:**\n1. What does the correction letter say? (Phase 1)\n2. What does state law require? (Phase 3A)\n3. Does the city add anything? (Phase 3.5)\n4. What's currently on the plan sheet? (Phase 3C)\n\n**Then categorize:**\n\n| Category | Meaning | Example |\n|----------|---------|---------|\n| `AUTO_FIXABLE` | Resolve by adding notes, marking checklists, updating labels | Missing CalGreen item, governing codes list |\n| `NEEDS_CONTRACTOR_INPUT` | Requires specific facts from the contractor | Sewer line size, finished grade elevations |\n| `NEEDS_PROFESSIONAL` | Requires licensed professional work (designer, engineer, HERS rater) | Structural calcs, fire-rated assembly detail |\n\n**Then generate questions** for `NEEDS_CONTRACTOR_INPUT` items. Each question includes `research_context` explaining why it's being asked and what the code requires. See `references\u002Foutput-schemas.md` for the `contractor_questions.json` schema.\n\n**Output files:** `corrections_categorized.json` + `contractor_questions.json`\n\n**Return `contractor_questions.json` to the UI.** This skill is now complete. Stop here.\n\n**What happens next:** The UI renders the questions. The contractor answers. Then the `adu-corrections-complete` skill takes the session directory + `contractor_answers.json` and generates the response package. That is a separate agent invocation — not a continuation of this one.\n\n## Timing\n\n| Phase | Time | Notes |\n|-------|------|-------|\n| Phase 1 | ~30 sec | Vision reading, 1-3 pages |\n| Phase 2 | ~90 sec | PDF extraction + manifest building |\n| Phase 3A | ~60 sec | Offline reference lookup |\n| Phase 3B | ~30 sec | City URL discovery (WebSearch only) |\n| Phase 3C | ~60 sec | Reading 5-8 PNGs |\n| Phase 3.5 | ~60-90 sec | City content extraction (WebFetch) |\n| Phase 3.5-fallback | ~2-3 min | Browser fallback (only if needed) |\n| Phase 4 | ~2 min | Merge + categorize + questions |\n| **Total (no fallback)** | **~4-5 min** | **Typical case** |\n| **Total (with fallback)** | **~6-8 min** | **Difficult city website** |\n\n## Important Notes\n\n- **This skill stops after Phase 4.** Do NOT generate response letters, professional scopes, or any Phase 5 outputs. That is the job of `adu-corrections-complete`.\n- **Research before questions.** Never generate contractor questions without first doing code research. The research makes the questions specific and actionable.\n- **Write high-quality research artifacts.** The `corrections_categorized.json` is the main handoff to the second skill. Every item must have its research context, code findings, and sheet observations fully documented — because the second skill runs cold with no conversation history.\n- **Sheet references are sacred.** Every sheet reference must come from `sheet-manifest.json`. Never guess.\n- **This tool helps contractors comply, not litigate.** Focus on *how to fix it*, not whether the correction is valid. If the city says fix it, help the contractor fix it.\n- **City research uses two passes.** Phase 3B (Discovery) runs fast via WebSearch in parallel with 3A\u002F3C. Phase 3.5 (Extraction) uses WebFetch against discovered URLs. Browser Fallback only runs if WebFetch has gaps. This two-pass approach cuts city research from ~5 min to ~90 sec for most cities.\n\n## References\n\n| File | Contents |\n|------|----------|\n| `references\u002Foutput-schemas.md` | JSON schemas for all output files — corrections_parsed, contractor_questions, contractor_answers |\n| `references\u002Fsubagent-prompts.md` | Full prompts for Phase 3 subagents (state law, city research, sheet viewer) + Phase 4 merge prompt |\n",{"name":31,"description":32},"adu-corrections-flow","Analyzes ADU permit corrections letters — the first half of the corrections pipeline. Reads the corrections letter, builds a sheet manifest from the plan binder, researches state and city codes, views referenced plan sheets, categorizes each correction item, and generates informed contractor questions. This skill should be used when a contractor receives a city corrections letter for an ADU permit. It coordinates three sub-skills (california-adu for state law, adu-city-research for city rules, adu-targeted-page-viewer for plan sheet navigation) to produce research artifacts and a UI-ready questions JSON. Does NOT generate the final response package — that is handled by adu-corrections-complete after the contractor answers questions. Triggers when a corrections letter PDF\u002FPNG is provided along with the plan binder PDF.","main","d1e2ecfb59900cb0b7e5999fe1c973add5c1f185",".claude\u002Fskills\u002Fecc\u002Fteam-builder","https:\u002F\u002Fgithub.com\u002Fsakamoto-family-smile\u002Fagent_monorepo","https:\u002F\u002Fgithub.com\u002Fsakamoto-family-smile","2026-11-08 20:29:14",{"source":8,"skillName":31,"commands":40,"prompt":50,"zipAvailable":51},[41,44,47],{"tool":42,"command":43},"npx","npx skills add sakamoto-family-smile\u002Fagent_monorepo --skill adu-corrections-flow",{"tool":45,"command":46},"bunx","bunx skills add sakamoto-family-smile\u002Fagent_monorepo --skill adu-corrections-flow",{"tool":48,"command":49},"pnpm","pnpm dlx skills add sakamoto-family-smile\u002Fagent_monorepo --skill adu-corrections-flow","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fsakamoto-family-smile\u002Fagent_monorepo\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fecc\u002Fteam-builder\u002FSKILL.md",true,[53,59],{"path":54,"resourceType":55,"resourceTypeDesc":56,"sizeBytes":57,"sourceHtmlUrl":58},"references\u002Foutput-schemas.md","REFERENCE","参考文档",12956,"https:\u002F\u002Fgithub.com\u002FMatthewLones\u002Fforgeflow\u002Fblob\u002Fmain\u002Fadu-corrections-flow\u002Freferences\u002Foutput-schemas.md",{"path":60,"resourceType":55,"resourceTypeDesc":56,"sizeBytes":61,"sourceHtmlUrl":62},"references\u002Fsubagent-prompts.md",18717,"https:\u002F\u002Fgithub.com\u002FMatthewLones\u002Fforgeflow\u002Fblob\u002Fmain\u002Fadu-corrections-flow\u002Freferences\u002Fsubagent-prompts.md",[64,98,106,126,149,178],{"skillNo":65,"name":66,"nameZh":6,"description":67,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":68,"categories":69,"keywords":78,"publishedAt":26},"01M6RW50THEFSKK42YNR8XFZA1","remotion-video-creation","Best practices for Remotion - Video creation in React. 29 domain-specific rules covering 3D, animations, audio, captions, charts, transitions, and more.","https:\u002F\u002Fgithub.com\u002Fsakamoto-family-smile\u002Fagent_monorepo\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fecc\u002Fremotion-video-creation\u002FSKILL.md",[70,75],{"categoryNo":71,"categoryName":72,"taxonomyType":73,"taxonomyTypeDesc":74},"01JBSEEDCAT000000000000009","前端开发","SCENARIO","使用场景",{"categoryNo":76,"categoryName":77,"taxonomyType":73,"taxonomyTypeDesc":74},"01JBSEEDCAT000000000000016","图像与多媒体",[79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,25],"Video","remotion video","remotion video creation","creation","Remotion","Charts","Captions","Audio","3D","Composition","Animation","best practices","more","animations","domain-specific","domain","three","Rules","Lottie",{"skillNo":99,"name":100,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":101,"categories":102,"keywords":103,"publishedAt":26},"01M6VCJ23NPECPSZJKGY691F75","hotcook-recipes","https:\u002F\u002Fgithub.com\u002Fsakamoto-family-smile\u002Fagent_monorepo\u002Fblob\u002Fmain\u002Fhotcook-agent\u002Fdata\u002Fskills\u002Fhotcook-recipes\u002FSKILL.md",[],[104,105],"hotcook recipes","recipes",{"skillNo":107,"name":108,"nameZh":6,"description":109,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":110,"categories":111,"keywords":120,"publishedAt":26},"01M6VCHHE00X5XHSDRD1KH1BW2","code-helper","Code review and assistance skill","https:\u002F\u002Fgithub.com\u002Fsakamoto-family-smile\u002Fagent_monorepo\u002Fblob\u002Fmain\u002Fagent-system-2\u002Fskills\u002Fcode-helper\u002Fskill.md",[112,115],{"categoryNo":113,"categoryName":114,"taxonomyType":73,"taxonomyTypeDesc":74},"01JBSEEDCAT000000000000007","代码审查",{"categoryNo":116,"categoryName":117,"taxonomyType":118,"taxonomyTypeDesc":119},"01JBSEEDCAT000000000000030","开发工程师","PROFESSION","适用职业",[121,122,123,124,125],"code helper","helper","code","code review","review",{"skillNo":127,"name":128,"nameZh":6,"description":129,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":130,"categories":131,"keywords":138,"publishedAt":26},"01M6VCH3Q31A20H6SQ1H0J7JB2","research-assistant","Web research assistant skill using Brave Search MCP","https:\u002F\u002Fgithub.com\u002Fsakamoto-family-smile\u002Fagent_monorepo\u002Fblob\u002Fmain\u002Fagent-system-1\u002Fskills\u002Fresearch-assistant\u002Fskill.md",[132,135],{"categoryNo":133,"categoryName":134,"taxonomyType":73,"taxonomyTypeDesc":74},"01JBSEEDCAT000000000000021","资料检索与研究",{"categoryNo":136,"categoryName":137,"taxonomyType":73,"taxonomyTypeDesc":74},"01JBSEEDCAT000000000000023","Skill与智能体开发",[139,140,141,142,143,144,145,146,147,148],"research assistant","research","assistant","MCP","web","search","brave search","web research","brave","web research assistant",{"skillNo":150,"name":151,"nameZh":6,"description":152,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":153,"categories":154,"keywords":159,"publishedAt":26},"01M6VCGZZGPNVJSCZ1JTMNZAHY","x-api","X\u002FTwitter API integration for posting tweets, threads, reading timelines, search, and analytics. Covers OAuth auth patterns, rate limits, and platform-native content posting. Use when the user wants to interact with X programmatically.","https:\u002F\u002Fgithub.com\u002Fsakamoto-family-smile\u002Fagent_monorepo\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fecc\u002Fx-api\u002FSKILL.md",[155,158],{"categoryNo":156,"categoryName":157,"taxonomyType":73,"taxonomyTypeDesc":74},"01JBSEEDCAT000000000000010","后端与接口",{"categoryNo":116,"categoryName":117,"taxonomyType":118,"taxonomyTypeDesc":119},[160,161,162,163,164,165,25,166,167,168,169,170,171,172,173,174,175,176,144,177],"x api","api","with","when","limits","rate","threads","reading","native","posting","platform","patterns","api integration","twitter","use","auth","analytics","OAuth",{"skillNo":179,"name":180,"nameZh":6,"description":181,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":182,"categories":183,"keywords":185,"publishedAt":26},"01M6VCGVX6DGB1FKTA53QBBCQB","workspace-surface-audit","Audit the active repo, MCP servers, plugins, connectors, env surfaces, and harness setup, then recommend the highest-value ECC-native skills, hooks, agents, and operator workflows. Use when the user wants help setting up Claude Code or understanding what capabilities are actually available in their environment.","https:\u002F\u002Fgithub.com\u002Fsakamoto-family-smile\u002Fagent_monorepo\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fecc\u002Fworkspace-surface-audit\u002FSKILL.md",[184],{"categoryNo":136,"categoryName":137,"taxonomyType":73,"taxonomyTypeDesc":74},[186,187,188,189,142,190,191,123,192,193,194,195,174,196,197,198,199,168,200,163],"workspace surface audit","workspace","audit","hooks","workflows","connectors","capabilities","harness","repo","environment","up","harness setup","active","env","value"]