@vibe-agent-toolkit/vat-example-cat-agents 0.2.0-rc.6 → 0.2.0-rc.8

package/dist/generated/resources/skills/SKILL.js

@@ -7,7 +7,7 @@ export const meta = {
   description: "Comprehensive orchestration guide for Claude Code using the vat-example-cat-agents toolkit"
 };
 
-export const text = "\n# VAT Example Cat Agents\n\nComprehensive orchestration guide for Claude Code using the vat-example-cat-agents toolkit.\n\n## Purpose: For Claude Code, Not the LLM\n\n**Key distinction:**\n- **This file** = Guidance for Claude Code (how to orchestrate agents)\n- **Agent resources** = Content for the LLM (what to say/know)\n\nThis skill references agent resources via markdown links but doesn\'t duplicate LLM prompts.\n\n## Agent Inventory\n\nThe vat-example-cat-agents package provides 8 agents across 4 archetypes:\n\n### Pure Function Tools (2 agents)\n- **haiku-validator** - Validates 5-7-5 syllable structure + kigo/kireji\n- **name-validator** - Quirky characteristic-based validation\n\n### One-Shot LLM Analyzers (4 agents)\n- **photo-analyzer** - Vision LLM extracts characteristics from images\n- **description-parser** - Text parsing extracts characteristics from descriptions\n- **name-generator** - Creates characteristic-based cat names\n- **haiku-generator** - Composes haikus about cats\n\n### Conversational Assistant (1 agent)\n- **breed-advisor** - Multi-turn breed selection through natural dialogue\n\n### External Event Integrator (1 agent)\n- **human-approval** - HITL approval gate (mockable)\n\n## When to Use This Skill\n\nTrigger this skill when:\n- User wants help selecting a cat breed\n- User has a cat photo and wants analysis\n- User needs a cat name suggestion\n- User wants a haiku about their cat\n- User needs validation of generated content (names, haikus)\n- User wants to combine multiple agents in a workflow\n\n## High-Level Orchestration Patterns\n\n### Pattern 1: Single Agent (Simple)\n\nUse when user has a straightforward request that maps to one agent.\n\n**Example workflows:**\n- \"What cat breed should I get?\" → [breed-advisor](../agents/breed-advisor.md)\n- \"Analyze this cat photo\" → [photo-analyzer](../agents/photo-analyzer.md)\n- \"Is this a valid haiku?\" → haiku-validator (pure function)\n\n### Pattern 2: Sequential Pipeline (Multi-Agent)\n\nUse when output of one agent feeds into another.\n\n**Example workflows:**\n1. Photo → Characteristics → Name\n   - [photo-analyzer](../agents/photo-analyzer.md) → [name-generator](../agents/name-generator.md)\n2. Photo → Characteristics → Haiku\n   - [photo-analyzer](../agents/photo-analyzer.md) → [haiku-generator](../agents/haiku-generator.md)\n3. Description → Characteristics → Name → Validation\n   - [description-parser](../agents/description-parser.md) → [name-generator](../agents/name-generator.md) → name-validator\n\n### Pattern 3: Generate-Validate Loop (Iterative)\n\nUse when generator produces content that needs validation with retry logic.\n\n**Example workflows:**\n1. Name generation with validation\n   - Generate → Validate → If invalid, retry with feedback\n   - Uses: [name-generator](../agents/name-generator.md) + name-validator\n2. Haiku generation with validation\n   - Generate → Validate → If invalid, retry\n   - Uses: [haiku-generator](../agents/haiku-generator.md) + haiku-validator\n\n**Orchestration tip:** Generator agents have NO knowledge of validation rules. This is intentional - forces iteration and tests feedback loops.\n\n### Pattern 4: HITL Approval Gate (External Event)\n\nUse when decision requires human judgment.\n\n**Example workflows:**\n1. Breed application approval\n   - Gather info → Generate application → [human-approval](../agents/human-approval.md) → Process result\n2. Name approval before finalization\n   - Generate names → Present options → Human selects → Finalize\n\n## Detailed Workflow Orchestration\n\n### Workflow A: Breed Selection (Conversational)\n\n**When:** User wants help finding the right cat breed.\n\n**Agent:** [breed-advisor](../agents/breed-advisor.md)\n\n**Orchestration strategy:**\n1. **Phase 1 - Gathering** (see [Conversation Strategy](../agents/breed-advisor.md#conversation-strategy))\n   - Collect ≥4 factors including music preference\n   - ONE question at a time (don\'t bombard)\n   - Extract factors after each turn\n   - Monitor readiness: \`factorsCollected >= 4 && musicPreference != null\`\n\n2. **Phase 2 - Recommendation**\n   - Present 3-5 matched breeds\n   - Allow exploration, questions, comparisons\n   - Use conversational formatting (not data dump)\n\n3. **Phase 3 - Selection**\n   - Detect selection signals (\"I\'ll take\", \"sounds good\")\n   - Conclude gracefully\n   - Provide next steps or exit instruction\n\n**Critical factor:** Music preference is the PRIMARY compatibility factor. Ask early, use as conversation anchor. See [Music Preference Insight](../agents/breed-advisor.md#music-preference-insight) for mappings.\n\n**Reference implementation:** See [cat-breed-selection.md](./cat-breed-selection.md) for detailed breed selection orchestration.\n\n### Workflow B: Photo Analysis Pipeline\n\n**When:** User provides a cat photo and wants analysis, name, or haiku.\n\n**Agents:** [photo-analyzer](../agents/photo-analyzer.md) → [name-generator](../agents/name-generator.md) OR [haiku-generator](../agents/haiku-generator.md)\n\n**Orchestration strategy:**\n\n\`\`\`\nStep 1: Analyze Photo\n- Input: Image path/URL\n- Agent: photo-analyzer\n- Output: CatCharacteristics\n- Note: Supports mock mode (reads EXIF) for testing\n\nStep 2: Generate Content\n- Input: CatCharacteristics from Step 1\n- Agent: name-generator OR haiku-generator\n- Output: NameSuggestion OR Haiku\n\nOptional Step 3: Validate\n- Input: Generated content + original characteristics\n- Agent: name-validator OR haiku-validator (pure functions)\n- Output: Validation result\n- If invalid: Retry Step 2 with feedback\n\`\`\`\n\n**Mockable behavior:** photo-analyzer reads EXIF metadata in mock mode. For production, set \`mockable: false\` to use real vision API.\n\n### Workflow C: Text Description Pipeline\n\n**When:** User describes a cat in text (no photo).\n\n**Agents:** [description-parser](../agents/description-parser.md) → [name-generator](../agents/name-generator.md) OR [haiku-generator](../agents/haiku-generator.md)\n\n**Orchestration strategy:**\n\n\`\`\`\nStep 1: Parse Description\n- Input: Text description\n- Agent: description-parser\n- Output: CatCharacteristics (same schema as photo-analyzer!)\n\nStep 2-3: Same as Workflow B\n- Multi-modal convergence: photo and text produce same schema\n- Downstream agents (name-gen, haiku-gen) work with either\n\`\`\`\n\n**Key insight:** Photo and text paths converge at \`CatCharacteristics\` schema. This enables multi-modal workflows without agent changes.\n\n### Workflow D: Generate-Validate-Retry Loop\n\n**When:** Generator produces content that must pass validation rules.\n\n**Pattern:** Generator (LLM) + Validator (pure function) + Retry logic\n\n**Implementation:**\n\n\`\`\`typescript\n// Pseudo-code for orchestration\nlet attempts = 0;\nconst maxAttempts = 3;\n\nwhile (attempts < maxAttempts) {\n  // Generate content\n  const suggestion = await generator(characteristics);\n\n  // Validate (pure function, instant)\n  const validation = validator(suggestion, characteristics);\n\n  if (validation.status === \'valid\') {\n    return suggestion; // Success!\n  }\n\n  // Retry with feedback\n  attempts++;\n  // Optional: Adjust approach based on validation.reason\n}\n\nthrow new Error(\'Could not generate valid content after 3 attempts\');\n\`\`\`\n\n**Why this pattern works:**\n- Generator has NO knowledge of validation rules (isolated concerns)\n- Validator is deterministic (pure function, fast)\n- Feedback loop tests multi-turn orchestration\n- ~60-70% initial rejection rate forces iteration\n\n**Agents using this pattern:**\n- [name-generator](../agents/name-generator.md) + name-validator\n- [haiku-generator](../agents/haiku-generator.md) + haiku-validator\n\n### Workflow E: HITL Approval Gate\n\n**When:** Decision requires human judgment (compliance, taste, ethics).\n\n**Agent:** [human-approval](../agents/human-approval.md)\n\n**Orchestration strategy:**\n\n\`\`\`\nStep 1: Prepare Request\n- Gather all necessary context\n- Format clearly for human reviewer\n- Include relevant data (characteristics, generated content, reasoning)\n\nStep 2: Emit Approval Request\n- Agent: human-approval\n- Input: Request payload\n- Behavior: Blocks waiting for response\n- Timeout: Configurable (default: 60s)\n\nStep 3: Handle Response\n- Approved: Continue workflow\n- Rejected: Handle gracefully (retry, inform user, etc.)\n- Timeout: Fall back to safe default or escalate\n\`\`\`\n\n**Mockable behavior:** Set \`mockable: true\` to auto-approve without human (useful for testing).\n\n**Real-world uses:**\n- Breeding application approval\n- Name selection from alternatives\n- Content moderation decisions\n\n## CLI Exposure for Pure Functions\n\nPure function tools (validators) can be exposed via CLI for direct invocation.\n\n### Usage Pattern\n\n\`\`\`bash\n# Pass JSON input, get JSON/YAML output\necho \'{\"name\": \"Mr. Whiskers\", \"characteristics\": {...}}\' | vat agents validate-name\n\n# Output format (using 2>&1):\n# [stdout - complete output]\n# ---\n# [stderr - error messages]\n\`\`\`\n\n### Implementation Requirements\n\n1. **Input:** JSON on stdin\n2. **Output:** JSON or YAML on stdout (complete, flushed)\n3. **Errors:** stderr (separated with \`---\` when using \`2>&1\`)\n4. **Exit codes:** 0 = success, 1 = validation failure, 2 = error\n\n### Sequence Example\n\n\`\`\`\n[Start]\n↓\nRead stdin (JSON input)\n↓\nParse and validate input schema\n↓\nExecute pure function\n↓\nFlush stdout with complete result\n↓\nPrint \"---\" separator\n↓\nFlush stderr with any error messages\n↓\nExit with appropriate code\n\`\`\`\n\n### Benefits\n\n- MCP can map to CLI calls (fast, stateless)\n- No long-running processes for pure functions\n- Clear separation of output vs errors\n- Composable with other CLI tools\n\n### Applicable Agents\n\n- **name-validator** - \`vat agents validate-name\`\n- **haiku-validator** - \`vat agents validate-haiku\`\n- Future: Any pure function tool\n\n## What Claude Code Does vs What Agents Do\n\n### Claude Code\'s Role (This Skill):\n\n**Orchestration:**\n- Select which agent(s) to use\n- Chain agents in workflows (pipelines)\n- Manage state between agents (pass outputs as inputs)\n- Handle retries and error recovery\n\n**Monitoring:**\n- Track conversation phase (gathering, recommendation, selection)\n- Count validation attempts (retry limits)\n- Detect completion signals (phase transitions)\n- Monitor timeouts (HITL approvals)\n\n**Decision Making:**\n- When to transition between phases\n- When to retry vs give up\n- Which agent path to take (photo vs text)\n- How to present results to user\n\n### Agents\' Role (Agent Resources):\n\n**Pure Functions:**\n- Execute deterministic logic (validation rules)\n- Return results instantly\n- No side effects, no state\n\n**LLM Analyzers:**\n- Extract structured data from unstructured input\n- Apply domain knowledge to classification\n- Generate creative content (names, haikus)\n\n**Conversational Assistants:**\n- Conduct natural dialogue\n- Accumulate context over turns\n- Make recommendations based on collected factors\n\n**Event Integrators:**\n- Emit events to external systems\n- Block waiting for responses\n- Handle timeouts and errors\n\n**Key insight:** Agents are specialized, focused on their domain. Claude Code provides the glue, orchestration, and workflow logic.\n\n## Content Separation Guidelines\n\n### What Belongs in Agent Resources (LLM-Facing):\n\n✅ System prompts and LLM instructions\n✅ Domain knowledge (breed database, syllable counting rules, kigo lists)\n✅ Extraction formats (JSON schemas, output templates)\n✅ Examples for few-shot learning\n✅ Natural language mappings\n✅ Validation rules and constraints\n\n**Location:** \`resources/agents/*.md\`\n\n### What Belongs in Skills (Claude Code-Facing):\n\n✅ When to trigger agents (user intent signals)\n✅ How to chain agents (workflow patterns)\n✅ What to monitor (readiness criteria, retry limits)\n✅ Debugging guidance (common issues, pitfalls)\n✅ Meta-strategy (why photo first, when to retry)\n✅ CLI exposure patterns\n\n**Location:** \`resources/skills/*.md\` (this file)\n\n### Overlap (Reference, Don\'t Duplicate):\n\n⚠️ Workflow structure (skill explains orchestration, agent implements)\n⚠️ Phase transitions (skill monitors, agent executes)\n⚠️ Validation criteria (skill manages retries, agent defines rules)\n\n**Resolution:** Keep agent resources authoritative. Skills REFERENCE via links, don\'t duplicate.\n\n## Common Pitfalls\n\n### ❌ Don\'t: Call Agents Without Understanding Their Archetype\n\nEach archetype has different behavior:\n- Pure functions: Instant, deterministic\n- LLM analyzers: Single call, non-deterministic\n- Conversational: Multi-turn, stateful\n- Event integrators: Blocking, timeout handling\n\n### ✅ Do: Match Orchestration to Archetype\n\n\`\`\`\nPure function → Call directly, no retry needed\nLLM analyzer → Call once, parse result\nConversational → Multi-turn loop with state\nEvent integrator → Emit, wait, handle timeout\n\`\`\`\n\n### ❌ Don\'t: Skip Validation in Generate-Validate Loops\n\nValidators exist for a reason. Don\'t bypass them or assume LLM output is always valid.\n\n### ✅ Do: Embrace the Feedback Loop\n\n\`\`\`\nGenerate → Validate → If invalid, retry with feedback\n\`\`\`\n\nThis pattern tests real-world orchestration where first attempts often fail.\n\n### ❌ Don\'t: Mix Mock and Production Modes Accidentally\n\nMock mode (EXIF metadata) is fast and free. Production mode (real APIs) is slow and expensive. Be explicit about which mode you\'re using.\n\n### ✅ Do: Use Mock Mode for Development, Production for Deployment\n\n\`\`\`typescript\n// Development/testing\nconst result = await analyzePhoto(path, { mockable: true });\n\n// Production\nconst result = await analyzePhoto(path, { mockable: false });\n\`\`\`\n\n### ❌ Don\'t: Bombard Users with Questions in Conversational Flows\n\nOne question at a time. Give users space to think and respond naturally.\n\n### ✅ Do: Guide Conversation Gently\n\n\`\`\`\nBad: \"Tell me your music, living space, activity level, grooming, family, and allergies\"\nGood: \"What\'s your favorite type of music?\" → [response] → \"Great! Tell me about your living space...\"\n\`\`\`\n\n## Debugging: When Things Go Wrong\n\n### Issue: Photo Analysis Fails\n\n**Symptoms:** Vision API errors, unexpected characteristics\n\n**Debug steps:**\n1. Check if image path is valid\n2. Verify image is actually a photo (not text, PDF, etc.)\n3. Try mock mode first: \`{ mockable: true }\`\n4. Check API key and rate limits for production mode\n5. Verify EXIF metadata format if using mock mode\n\n### Issue: Name Validation Always Fails\n\n**Symptoms:** Name generator can\'t produce valid names after 3+ attempts\n\n**Debug steps:**\n1. Check validation rules: name-validator has quirky requirements\n2. Review characteristics: Validation rules depend on cat traits\n3. Verify characteristics schema: All required fields present?\n4. Check feedback loop: Is validation.reason being used for retries?\n\n**Common cause:** ~60-70% rejection rate is EXPECTED. This forces iteration and tests feedback loops.\n\n### Issue: Haiku Validation Rejects Everything\n\n**Symptoms:** Haiku generator struggles to meet 5-7-5 syllable structure\n\n**Debug steps:**\n1. Check syllable counting algorithm: May differ from LLM\'s internal counting\n2. Verify kigo (seasonal word) presence: Required for valid haiku\n3. Check kireji (cutting word) detection: Optional but improves score\n4. Review haiku format: Must be exactly 3 lines\n\n**Solution:** Allow multiple retry attempts (3-5). Haiku generation is hard.\n\n### Issue: Conversation Stalls in Breed Selection\n\n**Symptoms:** Agent doesn\'t transition to recommendations\n\n**Debug steps:**\n1. Check factor count: Need ≥4 factors collected\n2. Verify music preference: REQUIRED for transition\n3. Review conversation history: Are factors being extracted?\n4. Check readiness criteria: Session state vs actual factors\n\n**Solution:** Ensure extraction happens after each turn. Monitor \`factorsCollected\` metadata.\n\n### Issue: HITL Approval Times Out\n\n**Symptoms:** human-approval agent returns timeout status\n\n**Debug steps:**\n1. Check timeout value: Default 60s, may need adjustment\n2. Verify event emission: Is request actually sent?\n3. Check mock mode: Set \`mockable: true\` for testing\n4. Review request format: Is it clear what needs approval?\n\n**Solution:** For testing, use mock mode. For production, increase timeout or add retry logic.\n\n## Agent Resource Links\n\n### Pure Function Tools\n- name-validator (no agent resource - pure TypeScript logic)\n- haiku-validator (no agent resource - pure TypeScript logic)\n\n### One-Shot LLM Analyzers\n- [Photo Analyzer](../agents/photo-analyzer.md) - Vision analysis\n- [Description Parser](../agents/description-parser.md) - Text parsing\n- [Name Generator](../agents/name-generator.md) - Creative naming\n- [Haiku Generator](../agents/haiku-generator.md) - Haiku composition\n\n### Conversational Assistant\n- [Breed Advisor](../agents/breed-advisor.md) - Multi-turn breed selection\n  - [Welcome Message](../agents/breed-advisor.md#welcome-message)\n  - [Music Preference Insight](../agents/breed-advisor.md#music-preference-insight)\n  - [Factor Definitions](../agents/breed-advisor.md#factor-definitions)\n  - [Conversation Strategy](../agents/breed-advisor.md#conversation-strategy)\n  - [Factor Extraction Prompt](../agents/breed-advisor.md#factor-extraction-prompt)\n  - [Transition Message](../agents/breed-advisor.md#transition-message)\n  - [Recommendation Presentation](../agents/breed-advisor.md#recommendation-presentation-prompt)\n  - [Selection Extraction](../agents/breed-advisor.md#selection-extraction-prompt)\n  - [Conclusion Prompt](../agents/breed-advisor.md#conclusion-prompt)\n\n### External Event Integrator\n- [Human Approval](../agents/human-approval.md) - HITL approval gate\n\n### Related Documentation\n- [Cat Breed Selection Skill](./cat-breed-selection.md) - Detailed breed advisor orchestration\n- [Package README](../../README.md) - Human-facing documentation\n- [Package Structure](../../docs/structure.md) - Technical organization\n\n## Success Criteria\n\nYour orchestration is successful when:\n\n**For Single Agents:**\n- Agent called with valid input schema\n- Output parsed and validated correctly\n- Errors handled gracefully\n- User receives clear, actionable results\n\n**For Pipelines:**\n- Data flows correctly between agents\n- Schema compatibility maintained (CatCharacteristics convergence)\n- Intermediate results stored appropriately\n- Final output meets user\'s original intent\n\n**For Loops:**\n- Retry logic prevents infinite loops (max attempts)\n- Feedback improves subsequent generations\n- User informed of progress (\"Generating... Attempt 2 of 3\")\n- Success achieved within reasonable attempts\n\n**For Conversational Flows:**\n- User feels heard, not interrogated\n- Natural dialogue rhythm maintained\n- Factors collected efficiently (4-6 turns typical)\n- Recommendations feel personalized\n- Selection confirmed clearly\n\n**For HITL Workflows:**\n- Request clearly formatted for human reviewer\n- Timeout handling prevents indefinite blocking\n- Approval/rejection handled appropriately\n- User understands what happened\n\n## Advanced Orchestration Patterns\n\n### Pattern: Parallel Execution\n\nWhen agents don\'t depend on each other, run them in parallel.\n\n**Example:** Generate both name and haiku from same characteristics\n\n\`\`\`typescript\nconst [name, haiku] = await Promise.all([\n  generateCatName(characteristics),\n  generateCatHaiku(characteristics),\n]);\n\`\`\`\n\n**Benefits:** Faster execution, better user experience\n\n### Pattern: Fallback Chain\n\nWhen one agent fails, try alternatives.\n\n**Example:** Photo analysis with text description fallback\n\n\`\`\`typescript\nlet characteristics;\ntry {\n  characteristics = await analyzePhoto(imagePath);\n} catch {\n  // Fallback to text description\n  const description = await getUserDescription();\n  characteristics = await parseDescription(description);\n}\n\`\`\`\n\n**Benefits:** Resilience, better error handling\n\n### Pattern: Conditional Routing\n\nChoose agent path based on user input type.\n\n**Example:** Multi-modal input handling\n\n\`\`\`typescript\nif (input.type === \'image\') {\n  characteristics = await analyzePhoto(input.imagePath);\n} else if (input.type === \'text\') {\n  characteristics = await parseDescription(input.description);\n} else {\n  // Conversational gathering\n  characteristics = await breedAdvisor.gather();\n}\n\`\`\`\n\n**Benefits:** Flexible input handling, better UX\n\n### Pattern: Staged Approval\n\nBreak complex workflows into approval stages.\n\n**Example:** Multi-stage breeding application\n\n\`\`\`typescript\n// Stage 1: Basic info approval\nconst basicApproval = await humanApproval({ stage: \'basic\', data });\nif (!basicApproval.approved) return;\n\n// Stage 2: Detailed questionnaire\nconst detailApproval = await humanApproval({ stage: \'detail\', data });\nif (!detailApproval.approved) return;\n\n// Stage 3: Final review\nconst finalApproval = await humanApproval({ stage: \'final\', data });\n\`\`\`\n\n**Benefits:** Checkpoints prevent wasted work, clearer decision points\n\n## CLI Integration Examples\n\n### Example 1: Validate Name via CLI\n\n\`\`\`bash\n# Input\necho \'{\n  \"name\": \"Mr. Whiskers\",\n  \"characteristics\": {\n    \"physical\": {\n      \"furColor\": \"Orange\",\n      \"furPattern\": \"Tabby\",\n      \"size\": \"medium\"\n    },\n    \"behavioral\": {\n      \"personality\": [\"Playful\", \"Curious\"]\n    }\n  }\n}\' | vat agents validate-name\n\n# Output (stdout)\n{\n  \"status\": \"valid\",\n  \"reason\": \"Name meets all quirky validation rules\",\n  \"confidence\": 0.95\n}\n---\n# (stderr - empty if no errors)\n\`\`\`\n\n### Example 2: Validate Haiku via CLI\n\n\`\`\`bash\n# Input\necho \'{\n  \"lines\": [\n    \"Orange sunset fur\",\n    \"Paws dance across the tatami\",\n    \"Zen master purrs\"\n  ]\n}\' | vat agents validate-haiku\n\n# Output (stdout)\n{\n  \"status\": \"valid\",\n  \"syllableCounts\": [5, 7, 5],\n  \"hasKigo\": true,\n  \"kigo\": \"sunset\",\n  \"hasKireji\": false,\n  \"errors\": []\n}\n---\n# (stderr - empty if no errors)\n\`\`\`\n\n### Example 3: Error Handling\n\n\`\`\`bash\n# Invalid input\necho \'{\"invalid\": \"schema\"}\' | vat agents validate-name\n\n# Output\n# (stdout - empty)\n---\nError: Invalid input schema. Expected \'name\' and \'characteristics\' fields.\nSchema validation failed:\n  - Missing required field: name\n  - Missing required field: characteristics\n# (exit code: 2)\n\`\`\`\n\n## Next Steps\n\nOnce you\'ve successfully orchestrated agents:\n\n1. **Explore Combinations** - Try chaining agents in new ways\n2. **Add Custom Validation** - Create your own quirky validation rules\n3. **Build New Workflows** - Combine agents for complex use cases\n4. **Contribute Patterns** - Share successful orchestration strategies\n5. **Package Skills** - Create distributable skill packages with pure functions\n\n## Questions?\n\n- **For orchestration patterns:** This file (SKILL.md)\n- **For agent-specific details:** See individual agent resources\n- **For implementation examples:** See examples/ directory\n- **For archetype theory:** See package README.md\n";
+export const text = "\n# VAT Example Cat Agents\n\nComprehensive orchestration guide for Claude Code using the vat-example-cat-agents toolkit.\n\n## Purpose: For Claude Code, Not the LLM\n\n**Key distinction:**\n- **This file** = Guidance for Claude Code (how to orchestrate agents)\n- **Agent resources** = Content for the LLM (what to say/know)\n\nThis skill references agent resources via markdown links but doesn\'t duplicate LLM prompts.\n\n## Agent Inventory\n\nThe vat-example-cat-agents package provides 8 agents across 4 archetypes:\n\n### Pure Function Tools (2 agents)\n- **haiku-validator** - Validates 5-7-5 syllable structure + kigo/kireji\n- **name-validator** - Quirky characteristic-based validation\n\n### One-Shot LLM Analyzers (4 agents)\n- **photo-analyzer** - Vision LLM extracts characteristics from images\n- **description-parser** - Text parsing extracts characteristics from descriptions\n- **name-generator** - Creates characteristic-based cat names\n- **haiku-generator** - Composes haikus about cats\n\n### Conversational Assistant (1 agent)\n- **breed-advisor** - Multi-turn breed selection through natural dialogue\n\n### External Event Integrator (1 agent)\n- **human-approval** - HITL approval gate (mockable)\n\n## When to Use This Skill\n\nTrigger this skill when:\n- User wants help selecting a cat breed\n- User has a cat photo and wants analysis\n- User needs a cat name suggestion\n- User wants a haiku about their cat\n- User needs validation of generated content (names, haikus)\n- User wants to combine multiple agents in a workflow\n\n## High-Level Orchestration Patterns\n\n### Pattern 1: Single Agent (Simple)\n\nUse when user has a straightforward request that maps to one agent.\n\n**Example workflows:**\n- \"What cat breed should I get?\" → [breed-advisor](../agents/breed-advisor.md)\n- \"Analyze this cat photo\" → [photo-analyzer](../agents/photo-analyzer.md)\n- \"Is this a valid haiku?\" → haiku-validator (pure function)\n\n### Pattern 2: Sequential Pipeline (Multi-Agent)\n\nUse when output of one agent feeds into another.\n\n**Example workflows:**\n1. Photo → Characteristics → Name\n   - [photo-analyzer](../agents/photo-analyzer.md) → [name-generator](../agents/name-generator.md)\n2. Photo → Characteristics → Haiku\n   - [photo-analyzer](../agents/photo-analyzer.md) → [haiku-generator](../agents/haiku-generator.md)\n3. Description → Characteristics → Name → Validation\n   - [description-parser](../agents/description-parser.md) → [name-generator](../agents/name-generator.md) → name-validator\n\n### Pattern 3: Generate-Validate Loop (Iterative)\n\nUse when generator produces content that needs validation with retry logic.\n\n**Example workflows:**\n1. Name generation with validation\n   - Generate → Validate → If invalid, retry with feedback\n   - Uses: [name-generator](../agents/name-generator.md) + name-validator\n2. Haiku generation with validation\n   - Generate → Validate → If invalid, retry\n   - Uses: [haiku-generator](../agents/haiku-generator.md) + haiku-validator\n\n**Orchestration tip:** Generator agents have NO knowledge of validation rules. This is intentional - forces iteration and tests feedback loops.\n\n### Pattern 4: HITL Approval Gate (External Event)\n\nUse when decision requires human judgment.\n\n**Example workflows:**\n1. Breed application approval\n   - Gather info → Generate application → [human-approval](../agents/human-approval.md) → Process result\n2. Name approval before finalization\n   - Generate names → Present options → Human selects → Finalize\n\n## Detailed Workflow Orchestration\n\nStep-by-step orchestration for each archetype — inputs, hand-offs and failure handling — is in\n[workflow-orchestration.md](./workflow-orchestration.md).\n\n## CLI Exposure for Pure Functions\n\nPure function tools (validators) can be exposed via CLI for direct invocation.\n\n### Usage Pattern\n\n\`\`\`bash\n# Pass JSON input, get JSON/YAML output\necho \'{\"name\": \"Mr. Whiskers\", \"characteristics\": {...}}\' | vat agents validate-name\n\n# Output format (using 2>&1):\n# [stdout - complete output]\n# ---\n# [stderr - error messages]\n\`\`\`\n\n### Implementation Requirements\n\n1. **Input:** JSON on stdin\n2. **Output:** JSON or YAML on stdout (complete, flushed)\n3. **Errors:** stderr (separated with \`---\` when using \`2>&1\`)\n4. **Exit codes:** 0 = success, 1 = validation failure, 2 = error\n\n### Sequence Example\n\n\`\`\`\n[Start]\n↓\nRead stdin (JSON input)\n↓\nParse and validate input schema\n↓\nExecute pure function\n↓\nFlush stdout with complete result\n↓\nPrint \"---\" separator\n↓\nFlush stderr with any error messages\n↓\nExit with appropriate code\n\`\`\`\n\n### Benefits\n\n- MCP can map to CLI calls (fast, stateless)\n- No long-running processes for pure functions\n- Clear separation of output vs errors\n- Composable with other CLI tools\n\n### Applicable Agents\n\n- **name-validator** - \`vat agents validate-name\`\n- **haiku-validator** - \`vat agents validate-haiku\`\n- Future: Any pure function tool\n\n## What Claude Code Does vs What Agents Do\n\n### Claude Code\'s Role (This Skill):\n\n**Orchestration:**\n- Select which agent(s) to use\n- Chain agents in workflows (pipelines)\n- Manage state between agents (pass outputs as inputs)\n- Handle retries and error recovery\n\n**Monitoring:**\n- Track conversation phase (gathering, recommendation, selection)\n- Count validation attempts (retry limits)\n- Detect completion signals (phase transitions)\n- Monitor timeouts (HITL approvals)\n\n**Decision Making:**\n- When to transition between phases\n- When to retry vs give up\n- Which agent path to take (photo vs text)\n- How to present results to user\n\n### Agents\' Role (Agent Resources):\n\n**Pure Functions:**\n- Execute deterministic logic (validation rules)\n- Return results instantly\n- No side effects, no state\n\n**LLM Analyzers:**\n- Extract structured data from unstructured input\n- Apply domain knowledge to classification\n- Generate creative content (names, haikus)\n\n**Conversational Assistants:**\n- Conduct natural dialogue\n- Accumulate context over turns\n- Make recommendations based on collected factors\n\n**Event Integrators:**\n- Emit events to external systems\n- Block waiting for responses\n- Handle timeouts and errors\n\n**Key insight:** Agents are specialized, focused on their domain. Claude Code provides the glue, orchestration, and workflow logic.\n\n## Content Separation Guidelines\n\n### What Belongs in Agent Resources (LLM-Facing):\n\n✅ System prompts and LLM instructions\n✅ Domain knowledge (breed database, syllable counting rules, kigo lists)\n✅ Extraction formats (JSON schemas, output templates)\n✅ Examples for few-shot learning\n✅ Natural language mappings\n✅ Validation rules and constraints\n\n**Location:** \`resources/agents/*.md\`\n\n### What Belongs in Skills (Claude Code-Facing):\n\n✅ When to trigger agents (user intent signals)\n✅ How to chain agents (workflow patterns)\n✅ What to monitor (readiness criteria, retry limits)\n✅ Debugging guidance (common issues, pitfalls)\n✅ Meta-strategy (why photo first, when to retry)\n✅ CLI exposure patterns\n\n**Location:** \`resources/skills/*.md\` (this file)\n\n### Overlap (Reference, Don\'t Duplicate):\n\n⚠️ Workflow structure (skill explains orchestration, agent implements)\n⚠️ Phase transitions (skill monitors, agent executes)\n⚠️ Validation criteria (skill manages retries, agent defines rules)\n\n**Resolution:** Keep agent resources authoritative. Skills REFERENCE via links, don\'t duplicate.\n\n## Common Pitfalls\n\n### ❌ Don\'t: Call Agents Without Understanding Their Archetype\n\nEach archetype has different behavior:\n- Pure functions: Instant, deterministic\n- LLM analyzers: Single call, non-deterministic\n- Conversational: Multi-turn, stateful\n- Event integrators: Blocking, timeout handling\n\n### ✅ Do: Match Orchestration to Archetype\n\n\`\`\`\nPure function → Call directly, no retry needed\nLLM analyzer → Call once, parse result\nConversational → Multi-turn loop with state\nEvent integrator → Emit, wait, handle timeout\n\`\`\`\n\n### ❌ Don\'t: Skip Validation in Generate-Validate Loops\n\nValidators exist for a reason. Don\'t bypass them or assume LLM output is always valid.\n\n### ✅ Do: Embrace the Feedback Loop\n\n\`\`\`\nGenerate → Validate → If invalid, retry with feedback\n\`\`\`\n\nThis pattern tests real-world orchestration where first attempts often fail.\n\n### ❌ Don\'t: Mix Mock and Production Modes Accidentally\n\nMock mode (EXIF metadata) is fast and free. Production mode (real APIs) is slow and expensive. Be explicit about which mode you\'re using.\n\n### ✅ Do: Use Mock Mode for Development, Production for Deployment\n\n\`\`\`typescript\n// Development/testing\nconst result = await analyzePhoto(path, { mockable: true });\n\n// Production\nconst result = await analyzePhoto(path, { mockable: false });\n\`\`\`\n\n### ❌ Don\'t: Bombard Users with Questions in Conversational Flows\n\nOne question at a time. Give users space to think and respond naturally.\n\n### ✅ Do: Guide Conversation Gently\n\n\`\`\`\nBad: \"Tell me your music, living space, activity level, grooming, family, and allergies\"\nGood: \"What\'s your favorite type of music?\" → [response] → \"Great! Tell me about your living space...\"\n\`\`\`\n\n## Debugging: When Things Go Wrong\n\n### Issue: Photo Analysis Fails\n\n**Symptoms:** Vision API errors, unexpected characteristics\n\n**Debug steps:**\n1. Check if image path is valid\n2. Verify image is actually a photo (not text, PDF, etc.)\n3. Try mock mode first: \`{ mockable: true }\`\n4. Check API key and rate limits for production mode\n5. Verify EXIF metadata format if using mock mode\n\n### Issue: Name Validation Always Fails\n\n**Symptoms:** Name generator can\'t produce valid names after 3+ attempts\n\n**Debug steps:**\n1. Check validation rules: name-validator has quirky requirements\n2. Review characteristics: Validation rules depend on cat traits\n3. Verify characteristics schema: All required fields present?\n4. Check feedback loop: Is validation.reason being used for retries?\n\n**Common cause:** ~60-70% rejection rate is EXPECTED. This forces iteration and tests feedback loops.\n\n### Issue: Haiku Validation Rejects Everything\n\n**Symptoms:** Haiku generator struggles to meet 5-7-5 syllable structure\n\n**Debug steps:**\n1. Check syllable counting algorithm: May differ from LLM\'s internal counting\n2. Verify kigo (seasonal word) presence: Required for valid haiku\n3. Check kireji (cutting word) detection: Optional but improves score\n4. Review haiku format: Must be exactly 3 lines\n\n**Solution:** Allow multiple retry attempts (3-5). Haiku generation is hard.\n\n### Issue: Conversation Stalls in Breed Selection\n\n**Symptoms:** Agent doesn\'t transition to recommendations\n\n**Debug steps:**\n1. Check factor count: Need ≥4 factors collected\n2. Verify music preference: REQUIRED for transition\n3. Review conversation history: Are factors being extracted?\n4. Check readiness criteria: Session state vs actual factors\n\n**Solution:** Ensure extraction happens after each turn. Monitor \`factorsCollected\` metadata.\n\n### Issue: HITL Approval Times Out\n\n**Symptoms:** human-approval agent returns timeout status\n\n**Debug steps:**\n1. Check timeout value: Default 60s, may need adjustment\n2. Verify event emission: Is request actually sent?\n3. Check mock mode: Set \`mockable: true\` for testing\n4. Review request format: Is it clear what needs approval?\n\n**Solution:** For testing, use mock mode. For production, increase timeout or add retry logic.\n\n## Agent Resource Links\n\n### Pure Function Tools\n- name-validator (no agent resource - pure TypeScript logic)\n- haiku-validator (no agent resource - pure TypeScript logic)\n\n### One-Shot LLM Analyzers\n- [Photo Analyzer](../agents/photo-analyzer.md) - Vision analysis\n- [Description Parser](../agents/description-parser.md) - Text parsing\n- [Name Generator](../agents/name-generator.md) - Creative naming\n- [Haiku Generator](../agents/haiku-generator.md) - Haiku composition\n\n### Conversational Assistant\n- [Breed Advisor](../agents/breed-advisor.md) - Multi-turn breed selection\n  - [Welcome Message](../agents/breed-advisor.md#welcome-message)\n  - [Music Preference Insight](../agents/breed-advisor.md#music-preference-insight)\n  - [Factor Definitions](../agents/breed-advisor.md#factor-definitions)\n  - [Conversation Strategy](../agents/breed-advisor.md#conversation-strategy)\n  - [Factor Extraction Prompt](../agents/breed-advisor.md#factor-extraction-prompt)\n  - [Transition Message](../agents/breed-advisor.md#transition-message)\n  - [Recommendation Presentation](../agents/breed-advisor.md#recommendation-presentation-prompt)\n  - [Selection Extraction](../agents/breed-advisor.md#selection-extraction-prompt)\n  - [Conclusion Prompt](../agents/breed-advisor.md#conclusion-prompt)\n\n### External Event Integrator\n- [Human Approval](../agents/human-approval.md) - HITL approval gate\n\n### Related Documentation\n- [Cat Breed Selection Skill](./cat-breed-selection.md) - Detailed breed advisor orchestration\n\n## Success Criteria\n\nYour orchestration is successful when:\n\n**For Single Agents:**\n- Agent called with valid input schema\n- Output parsed and validated correctly\n- Errors handled gracefully\n- User receives clear, actionable results\n\n**For Pipelines:**\n- Data flows correctly between agents\n- Schema compatibility maintained (CatCharacteristics convergence)\n- Intermediate results stored appropriately\n- Final output meets user\'s original intent\n\n**For Loops:**\n- Retry logic prevents infinite loops (max attempts)\n- Feedback improves subsequent generations\n- User informed of progress (\"Generating... Attempt 2 of 3\")\n- Success achieved within reasonable attempts\n\n**For Conversational Flows:**\n- User feels heard, not interrogated\n- Natural dialogue rhythm maintained\n- Factors collected efficiently (4-6 turns typical)\n- Recommendations feel personalized\n- Selection confirmed clearly\n\n**For HITL Workflows:**\n- Request clearly formatted for human reviewer\n- Timeout handling prevents indefinite blocking\n- Approval/rejection handled appropriately\n- User understands what happened\n\n## Advanced Patterns and CLI Integration\n\nParallel, conditional and checkpointed orchestration, plus worked CLI request/response pairs,\nare in [orchestration-patterns.md](./orchestration-patterns.md).\n";
 
 export const fragments = {
   purposeForClaudeCodeNotTheLlm: {
@@ -32,8 +32,8 @@ export const fragments = {
   },
   detailedWorkflowOrchestration: {
     header: "## Detailed Workflow Orchestration",
-    body: "### Workflow A: Breed Selection (Conversational)\n\n**When:** User wants help finding the right cat breed.\n\n**Agent:** [breed-advisor](../agents/breed-advisor.md)\n\n**Orchestration strategy:**\n1. **Phase 1 - Gathering** (see [Conversation Strategy](../agents/breed-advisor.md#conversation-strategy))\n   - Collect ≥4 factors including music preference\n   - ONE question at a time (don\'t bombard)\n   - Extract factors after each turn\n   - Monitor readiness: \`factorsCollected >= 4 && musicPreference != null\`\n\n2. **Phase 2 - Recommendation**\n   - Present 3-5 matched breeds\n   - Allow exploration, questions, comparisons\n   - Use conversational formatting (not data dump)\n\n3. **Phase 3 - Selection**\n   - Detect selection signals (\"I\'ll take\", \"sounds good\")\n   - Conclude gracefully\n   - Provide next steps or exit instruction\n\n**Critical factor:** Music preference is the PRIMARY compatibility factor. Ask early, use as conversation anchor. See [Music Preference Insight](../agents/breed-advisor.md#music-preference-insight) for mappings.\n\n**Reference implementation:** See [cat-breed-selection.md](./cat-breed-selection.md) for detailed breed selection orchestration.\n\n### Workflow B: Photo Analysis Pipeline\n\n**When:** User provides a cat photo and wants analysis, name, or haiku.\n\n**Agents:** [photo-analyzer](../agents/photo-analyzer.md) → [name-generator](../agents/name-generator.md) OR [haiku-generator](../agents/haiku-generator.md)\n\n**Orchestration strategy:**\n\n\`\`\`\nStep 1: Analyze Photo\n- Input: Image path/URL\n- Agent: photo-analyzer\n- Output: CatCharacteristics\n- Note: Supports mock mode (reads EXIF) for testing\n\nStep 2: Generate Content\n- Input: CatCharacteristics from Step 1\n- Agent: name-generator OR haiku-generator\n- Output: NameSuggestion OR Haiku\n\nOptional Step 3: Validate\n- Input: Generated content + original characteristics\n- Agent: name-validator OR haiku-validator (pure functions)\n- Output: Validation result\n- If invalid: Retry Step 2 with feedback\n\`\`\`\n\n**Mockable behavior:** photo-analyzer reads EXIF metadata in mock mode. For production, set \`mockable: false\` to use real vision API.\n\n### Workflow C: Text Description Pipeline\n\n**When:** User describes a cat in text (no photo).\n\n**Agents:** [description-parser](../agents/description-parser.md) → [name-generator](../agents/name-generator.md) OR [haiku-generator](../agents/haiku-generator.md)\n\n**Orchestration strategy:**\n\n\`\`\`\nStep 1: Parse Description\n- Input: Text description\n- Agent: description-parser\n- Output: CatCharacteristics (same schema as photo-analyzer!)\n\nStep 2-3: Same as Workflow B\n- Multi-modal convergence: photo and text produce same schema\n- Downstream agents (name-gen, haiku-gen) work with either\n\`\`\`\n\n**Key insight:** Photo and text paths converge at \`CatCharacteristics\` schema. This enables multi-modal workflows without agent changes.\n\n### Workflow D: Generate-Validate-Retry Loop\n\n**When:** Generator produces content that must pass validation rules.\n\n**Pattern:** Generator (LLM) + Validator (pure function) + Retry logic\n\n**Implementation:**\n\n\`\`\`typescript\n// Pseudo-code for orchestration\nlet attempts = 0;\nconst maxAttempts = 3;\n\nwhile (attempts < maxAttempts) {\n  // Generate content\n  const suggestion = await generator(characteristics);\n\n  // Validate (pure function, instant)\n  const validation = validator(suggestion, characteristics);\n\n  if (validation.status === \'valid\') {\n    return suggestion; // Success!\n  }\n\n  // Retry with feedback\n  attempts++;\n  // Optional: Adjust approach based on validation.reason\n}\n\nthrow new Error(\'Could not generate valid content after 3 attempts\');\n\`\`\`\n\n**Why this pattern works:**\n- Generator has NO knowledge of validation rules (isolated concerns)\n- Validator is deterministic (pure function, fast)\n- Feedback loop tests multi-turn orchestration\n- ~60-70% initial rejection rate forces iteration\n\n**Agents using this pattern:**\n- [name-generator](../agents/name-generator.md) + name-validator\n- [haiku-generator](../agents/haiku-generator.md) + haiku-validator\n\n### Workflow E: HITL Approval Gate\n\n**When:** Decision requires human judgment (compliance, taste, ethics).\n\n**Agent:** [human-approval](../agents/human-approval.md)\n\n**Orchestration strategy:**\n\n\`\`\`\nStep 1: Prepare Request\n- Gather all necessary context\n- Format clearly for human reviewer\n- Include relevant data (characteristics, generated content, reasoning)\n\nStep 2: Emit Approval Request\n- Agent: human-approval\n- Input: Request payload\n- Behavior: Blocks waiting for response\n- Timeout: Configurable (default: 60s)\n\nStep 3: Handle Response\n- Approved: Continue workflow\n- Rejected: Handle gracefully (retry, inform user, etc.)\n- Timeout: Fall back to safe default or escalate\n\`\`\`\n\n**Mockable behavior:** Set \`mockable: true\` to auto-approve without human (useful for testing).\n\n**Real-world uses:**\n- Breeding application approval\n- Name selection from alternatives\n- Content moderation decisions",
-    text: "## Detailed Workflow Orchestration\n\n### Workflow A: Breed Selection (Conversational)\n\n**When:** User wants help finding the right cat breed.\n\n**Agent:** [breed-advisor](../agents/breed-advisor.md)\n\n**Orchestration strategy:**\n1. **Phase 1 - Gathering** (see [Conversation Strategy](../agents/breed-advisor.md#conversation-strategy))\n   - Collect ≥4 factors including music preference\n   - ONE question at a time (don\'t bombard)\n   - Extract factors after each turn\n   - Monitor readiness: \`factorsCollected >= 4 && musicPreference != null\`\n\n2. **Phase 2 - Recommendation**\n   - Present 3-5 matched breeds\n   - Allow exploration, questions, comparisons\n   - Use conversational formatting (not data dump)\n\n3. **Phase 3 - Selection**\n   - Detect selection signals (\"I\'ll take\", \"sounds good\")\n   - Conclude gracefully\n   - Provide next steps or exit instruction\n\n**Critical factor:** Music preference is the PRIMARY compatibility factor. Ask early, use as conversation anchor. See [Music Preference Insight](../agents/breed-advisor.md#music-preference-insight) for mappings.\n\n**Reference implementation:** See [cat-breed-selection.md](./cat-breed-selection.md) for detailed breed selection orchestration.\n\n### Workflow B: Photo Analysis Pipeline\n\n**When:** User provides a cat photo and wants analysis, name, or haiku.\n\n**Agents:** [photo-analyzer](../agents/photo-analyzer.md) → [name-generator](../agents/name-generator.md) OR [haiku-generator](../agents/haiku-generator.md)\n\n**Orchestration strategy:**\n\n\`\`\`\nStep 1: Analyze Photo\n- Input: Image path/URL\n- Agent: photo-analyzer\n- Output: CatCharacteristics\n- Note: Supports mock mode (reads EXIF) for testing\n\nStep 2: Generate Content\n- Input: CatCharacteristics from Step 1\n- Agent: name-generator OR haiku-generator\n- Output: NameSuggestion OR Haiku\n\nOptional Step 3: Validate\n- Input: Generated content + original characteristics\n- Agent: name-validator OR haiku-validator (pure functions)\n- Output: Validation result\n- If invalid: Retry Step 2 with feedback\n\`\`\`\n\n**Mockable behavior:** photo-analyzer reads EXIF metadata in mock mode. For production, set \`mockable: false\` to use real vision API.\n\n### Workflow C: Text Description Pipeline\n\n**When:** User describes a cat in text (no photo).\n\n**Agents:** [description-parser](../agents/description-parser.md) → [name-generator](../agents/name-generator.md) OR [haiku-generator](../agents/haiku-generator.md)\n\n**Orchestration strategy:**\n\n\`\`\`\nStep 1: Parse Description\n- Input: Text description\n- Agent: description-parser\n- Output: CatCharacteristics (same schema as photo-analyzer!)\n\nStep 2-3: Same as Workflow B\n- Multi-modal convergence: photo and text produce same schema\n- Downstream agents (name-gen, haiku-gen) work with either\n\`\`\`\n\n**Key insight:** Photo and text paths converge at \`CatCharacteristics\` schema. This enables multi-modal workflows without agent changes.\n\n### Workflow D: Generate-Validate-Retry Loop\n\n**When:** Generator produces content that must pass validation rules.\n\n**Pattern:** Generator (LLM) + Validator (pure function) + Retry logic\n\n**Implementation:**\n\n\`\`\`typescript\n// Pseudo-code for orchestration\nlet attempts = 0;\nconst maxAttempts = 3;\n\nwhile (attempts < maxAttempts) {\n  // Generate content\n  const suggestion = await generator(characteristics);\n\n  // Validate (pure function, instant)\n  const validation = validator(suggestion, characteristics);\n\n  if (validation.status === \'valid\') {\n    return suggestion; // Success!\n  }\n\n  // Retry with feedback\n  attempts++;\n  // Optional: Adjust approach based on validation.reason\n}\n\nthrow new Error(\'Could not generate valid content after 3 attempts\');\n\`\`\`\n\n**Why this pattern works:**\n- Generator has NO knowledge of validation rules (isolated concerns)\n- Validator is deterministic (pure function, fast)\n- Feedback loop tests multi-turn orchestration\n- ~60-70% initial rejection rate forces iteration\n\n**Agents using this pattern:**\n- [name-generator](../agents/name-generator.md) + name-validator\n- [haiku-generator](../agents/haiku-generator.md) + haiku-validator\n\n### Workflow E: HITL Approval Gate\n\n**When:** Decision requires human judgment (compliance, taste, ethics).\n\n**Agent:** [human-approval](../agents/human-approval.md)\n\n**Orchestration strategy:**\n\n\`\`\`\nStep 1: Prepare Request\n- Gather all necessary context\n- Format clearly for human reviewer\n- Include relevant data (characteristics, generated content, reasoning)\n\nStep 2: Emit Approval Request\n- Agent: human-approval\n- Input: Request payload\n- Behavior: Blocks waiting for response\n- Timeout: Configurable (default: 60s)\n\nStep 3: Handle Response\n- Approved: Continue workflow\n- Rejected: Handle gracefully (retry, inform user, etc.)\n- Timeout: Fall back to safe default or escalate\n\`\`\`\n\n**Mockable behavior:** Set \`mockable: true\` to auto-approve without human (useful for testing).\n\n**Real-world uses:**\n- Breeding application approval\n- Name selection from alternatives\n- Content moderation decisions"
+    body: "Step-by-step orchestration for each archetype — inputs, hand-offs and failure handling — is in\n[workflow-orchestration.md](./workflow-orchestration.md).",
+    text: "## Detailed Workflow Orchestration\n\nStep-by-step orchestration for each archetype — inputs, hand-offs and failure handling — is in\n[workflow-orchestration.md](./workflow-orchestration.md)."
   },
   cliExposureForPureFunctions: {
     header: "## CLI Exposure for Pure Functions",
@@ -62,32 +62,17 @@ export const fragments = {
   },
   agentResourceLinks: {
     header: "## Agent Resource Links",
-    body: "### Pure Function Tools\n- name-validator (no agent resource - pure TypeScript logic)\n- haiku-validator (no agent resource - pure TypeScript logic)\n\n### One-Shot LLM Analyzers\n- [Photo Analyzer](../agents/photo-analyzer.md) - Vision analysis\n- [Description Parser](../agents/description-parser.md) - Text parsing\n- [Name Generator](../agents/name-generator.md) - Creative naming\n- [Haiku Generator](../agents/haiku-generator.md) - Haiku composition\n\n### Conversational Assistant\n- [Breed Advisor](../agents/breed-advisor.md) - Multi-turn breed selection\n  - [Welcome Message](../agents/breed-advisor.md#welcome-message)\n  - [Music Preference Insight](../agents/breed-advisor.md#music-preference-insight)\n  - [Factor Definitions](../agents/breed-advisor.md#factor-definitions)\n  - [Conversation Strategy](../agents/breed-advisor.md#conversation-strategy)\n  - [Factor Extraction Prompt](../agents/breed-advisor.md#factor-extraction-prompt)\n  - [Transition Message](../agents/breed-advisor.md#transition-message)\n  - [Recommendation Presentation](../agents/breed-advisor.md#recommendation-presentation-prompt)\n  - [Selection Extraction](../agents/breed-advisor.md#selection-extraction-prompt)\n  - [Conclusion Prompt](../agents/breed-advisor.md#conclusion-prompt)\n\n### External Event Integrator\n- [Human Approval](../agents/human-approval.md) - HITL approval gate\n\n### Related Documentation\n- [Cat Breed Selection Skill](./cat-breed-selection.md) - Detailed breed advisor orchestration\n- [Package README](../../README.md) - Human-facing documentation\n- [Package Structure](../../docs/structure.md) - Technical organization",
-    text: "## Agent Resource Links\n\n### Pure Function Tools\n- name-validator (no agent resource - pure TypeScript logic)\n- haiku-validator (no agent resource - pure TypeScript logic)\n\n### One-Shot LLM Analyzers\n- [Photo Analyzer](../agents/photo-analyzer.md) - Vision analysis\n- [Description Parser](../agents/description-parser.md) - Text parsing\n- [Name Generator](../agents/name-generator.md) - Creative naming\n- [Haiku Generator](../agents/haiku-generator.md) - Haiku composition\n\n### Conversational Assistant\n- [Breed Advisor](../agents/breed-advisor.md) - Multi-turn breed selection\n  - [Welcome Message](../agents/breed-advisor.md#welcome-message)\n  - [Music Preference Insight](../agents/breed-advisor.md#music-preference-insight)\n  - [Factor Definitions](../agents/breed-advisor.md#factor-definitions)\n  - [Conversation Strategy](../agents/breed-advisor.md#conversation-strategy)\n  - [Factor Extraction Prompt](../agents/breed-advisor.md#factor-extraction-prompt)\n  - [Transition Message](../agents/breed-advisor.md#transition-message)\n  - [Recommendation Presentation](../agents/breed-advisor.md#recommendation-presentation-prompt)\n  - [Selection Extraction](../agents/breed-advisor.md#selection-extraction-prompt)\n  - [Conclusion Prompt](../agents/breed-advisor.md#conclusion-prompt)\n\n### External Event Integrator\n- [Human Approval](../agents/human-approval.md) - HITL approval gate\n\n### Related Documentation\n- [Cat Breed Selection Skill](./cat-breed-selection.md) - Detailed breed advisor orchestration\n- [Package README](../../README.md) - Human-facing documentation\n- [Package Structure](../../docs/structure.md) - Technical organization"
+    body: "### Pure Function Tools\n- name-validator (no agent resource - pure TypeScript logic)\n- haiku-validator (no agent resource - pure TypeScript logic)\n\n### One-Shot LLM Analyzers\n- [Photo Analyzer](../agents/photo-analyzer.md) - Vision analysis\n- [Description Parser](../agents/description-parser.md) - Text parsing\n- [Name Generator](../agents/name-generator.md) - Creative naming\n- [Haiku Generator](../agents/haiku-generator.md) - Haiku composition\n\n### Conversational Assistant\n- [Breed Advisor](../agents/breed-advisor.md) - Multi-turn breed selection\n  - [Welcome Message](../agents/breed-advisor.md#welcome-message)\n  - [Music Preference Insight](../agents/breed-advisor.md#music-preference-insight)\n  - [Factor Definitions](../agents/breed-advisor.md#factor-definitions)\n  - [Conversation Strategy](../agents/breed-advisor.md#conversation-strategy)\n  - [Factor Extraction Prompt](../agents/breed-advisor.md#factor-extraction-prompt)\n  - [Transition Message](../agents/breed-advisor.md#transition-message)\n  - [Recommendation Presentation](../agents/breed-advisor.md#recommendation-presentation-prompt)\n  - [Selection Extraction](../agents/breed-advisor.md#selection-extraction-prompt)\n  - [Conclusion Prompt](../agents/breed-advisor.md#conclusion-prompt)\n\n### External Event Integrator\n- [Human Approval](../agents/human-approval.md) - HITL approval gate\n\n### Related Documentation\n- [Cat Breed Selection Skill](./cat-breed-selection.md) - Detailed breed advisor orchestration",
+    text: "## Agent Resource Links\n\n### Pure Function Tools\n- name-validator (no agent resource - pure TypeScript logic)\n- haiku-validator (no agent resource - pure TypeScript logic)\n\n### One-Shot LLM Analyzers\n- [Photo Analyzer](../agents/photo-analyzer.md) - Vision analysis\n- [Description Parser](../agents/description-parser.md) - Text parsing\n- [Name Generator](../agents/name-generator.md) - Creative naming\n- [Haiku Generator](../agents/haiku-generator.md) - Haiku composition\n\n### Conversational Assistant\n- [Breed Advisor](../agents/breed-advisor.md) - Multi-turn breed selection\n  - [Welcome Message](../agents/breed-advisor.md#welcome-message)\n  - [Music Preference Insight](../agents/breed-advisor.md#music-preference-insight)\n  - [Factor Definitions](../agents/breed-advisor.md#factor-definitions)\n  - [Conversation Strategy](../agents/breed-advisor.md#conversation-strategy)\n  - [Factor Extraction Prompt](../agents/breed-advisor.md#factor-extraction-prompt)\n  - [Transition Message](../agents/breed-advisor.md#transition-message)\n  - [Recommendation Presentation](../agents/breed-advisor.md#recommendation-presentation-prompt)\n  - [Selection Extraction](../agents/breed-advisor.md#selection-extraction-prompt)\n  - [Conclusion Prompt](../agents/breed-advisor.md#conclusion-prompt)\n\n### External Event Integrator\n- [Human Approval](../agents/human-approval.md) - HITL approval gate\n\n### Related Documentation\n- [Cat Breed Selection Skill](./cat-breed-selection.md) - Detailed breed advisor orchestration"
   },
   successCriteria: {
     header: "## Success Criteria",
     body: "Your orchestration is successful when:\n\n**For Single Agents:**\n- Agent called with valid input schema\n- Output parsed and validated correctly\n- Errors handled gracefully\n- User receives clear, actionable results\n\n**For Pipelines:**\n- Data flows correctly between agents\n- Schema compatibility maintained (CatCharacteristics convergence)\n- Intermediate results stored appropriately\n- Final output meets user\'s original intent\n\n**For Loops:**\n- Retry logic prevents infinite loops (max attempts)\n- Feedback improves subsequent generations\n- User informed of progress (\"Generating... Attempt 2 of 3\")\n- Success achieved within reasonable attempts\n\n**For Conversational Flows:**\n- User feels heard, not interrogated\n- Natural dialogue rhythm maintained\n- Factors collected efficiently (4-6 turns typical)\n- Recommendations feel personalized\n- Selection confirmed clearly\n\n**For HITL Workflows:**\n- Request clearly formatted for human reviewer\n- Timeout handling prevents indefinite blocking\n- Approval/rejection handled appropriately\n- User understands what happened",
     text: "## Success Criteria\n\nYour orchestration is successful when:\n\n**For Single Agents:**\n- Agent called with valid input schema\n- Output parsed and validated correctly\n- Errors handled gracefully\n- User receives clear, actionable results\n\n**For Pipelines:**\n- Data flows correctly between agents\n- Schema compatibility maintained (CatCharacteristics convergence)\n- Intermediate results stored appropriately\n- Final output meets user\'s original intent\n\n**For Loops:**\n- Retry logic prevents infinite loops (max attempts)\n- Feedback improves subsequent generations\n- User informed of progress (\"Generating... Attempt 2 of 3\")\n- Success achieved within reasonable attempts\n\n**For Conversational Flows:**\n- User feels heard, not interrogated\n- Natural dialogue rhythm maintained\n- Factors collected efficiently (4-6 turns typical)\n- Recommendations feel personalized\n- Selection confirmed clearly\n\n**For HITL Workflows:**\n- Request clearly formatted for human reviewer\n- Timeout handling prevents indefinite blocking\n- Approval/rejection handled appropriately\n- User understands what happened"
   },
-  advancedOrchestrationPatterns: {
-    header: "## Advanced Orchestration Patterns",
-    body: "### Pattern: Parallel Execution\n\nWhen agents don\'t depend on each other, run them in parallel.\n\n**Example:** Generate both name and haiku from same characteristics\n\n\`\`\`typescript\nconst [name, haiku] = await Promise.all([\n  generateCatName(characteristics),\n  generateCatHaiku(characteristics),\n]);\n\`\`\`\n\n**Benefits:** Faster execution, better user experience\n\n### Pattern: Fallback Chain\n\nWhen one agent fails, try alternatives.\n\n**Example:** Photo analysis with text description fallback\n\n\`\`\`typescript\nlet characteristics;\ntry {\n  characteristics = await analyzePhoto(imagePath);\n} catch {\n  // Fallback to text description\n  const description = await getUserDescription();\n  characteristics = await parseDescription(description);\n}\n\`\`\`\n\n**Benefits:** Resilience, better error handling\n\n### Pattern: Conditional Routing\n\nChoose agent path based on user input type.\n\n**Example:** Multi-modal input handling\n\n\`\`\`typescript\nif (input.type === \'image\') {\n  characteristics = await analyzePhoto(input.imagePath);\n} else if (input.type === \'text\') {\n  characteristics = await parseDescription(input.description);\n} else {\n  // Conversational gathering\n  characteristics = await breedAdvisor.gather();\n}\n\`\`\`\n\n**Benefits:** Flexible input handling, better UX\n\n### Pattern: Staged Approval\n\nBreak complex workflows into approval stages.\n\n**Example:** Multi-stage breeding application\n\n\`\`\`typescript\n// Stage 1: Basic info approval\nconst basicApproval = await humanApproval({ stage: \'basic\', data });\nif (!basicApproval.approved) return;\n\n// Stage 2: Detailed questionnaire\nconst detailApproval = await humanApproval({ stage: \'detail\', data });\nif (!detailApproval.approved) return;\n\n// Stage 3: Final review\nconst finalApproval = await humanApproval({ stage: \'final\', data });\n\`\`\`\n\n**Benefits:** Checkpoints prevent wasted work, clearer decision points",
-    text: "## Advanced Orchestration Patterns\n\n### Pattern: Parallel Execution\n\nWhen agents don\'t depend on each other, run them in parallel.\n\n**Example:** Generate both name and haiku from same characteristics\n\n\`\`\`typescript\nconst [name, haiku] = await Promise.all([\n  generateCatName(characteristics),\n  generateCatHaiku(characteristics),\n]);\n\`\`\`\n\n**Benefits:** Faster execution, better user experience\n\n### Pattern: Fallback Chain\n\nWhen one agent fails, try alternatives.\n\n**Example:** Photo analysis with text description fallback\n\n\`\`\`typescript\nlet characteristics;\ntry {\n  characteristics = await analyzePhoto(imagePath);\n} catch {\n  // Fallback to text description\n  const description = await getUserDescription();\n  characteristics = await parseDescription(description);\n}\n\`\`\`\n\n**Benefits:** Resilience, better error handling\n\n### Pattern: Conditional Routing\n\nChoose agent path based on user input type.\n\n**Example:** Multi-modal input handling\n\n\`\`\`typescript\nif (input.type === \'image\') {\n  characteristics = await analyzePhoto(input.imagePath);\n} else if (input.type === \'text\') {\n  characteristics = await parseDescription(input.description);\n} else {\n  // Conversational gathering\n  characteristics = await breedAdvisor.gather();\n}\n\`\`\`\n\n**Benefits:** Flexible input handling, better UX\n\n### Pattern: Staged Approval\n\nBreak complex workflows into approval stages.\n\n**Example:** Multi-stage breeding application\n\n\`\`\`typescript\n// Stage 1: Basic info approval\nconst basicApproval = await humanApproval({ stage: \'basic\', data });\nif (!basicApproval.approved) return;\n\n// Stage 2: Detailed questionnaire\nconst detailApproval = await humanApproval({ stage: \'detail\', data });\nif (!detailApproval.approved) return;\n\n// Stage 3: Final review\nconst finalApproval = await humanApproval({ stage: \'final\', data });\n\`\`\`\n\n**Benefits:** Checkpoints prevent wasted work, clearer decision points"
-  },
-  cliIntegrationExamples: {
-    header: "## CLI Integration Examples",
-    body: "### Example 1: Validate Name via CLI\n\n\`\`\`bash\n# Input\necho \'{\n  \"name\": \"Mr. Whiskers\",\n  \"characteristics\": {\n    \"physical\": {\n      \"furColor\": \"Orange\",\n      \"furPattern\": \"Tabby\",\n      \"size\": \"medium\"\n    },\n    \"behavioral\": {\n      \"personality\": [\"Playful\", \"Curious\"]\n    }\n  }\n}\' | vat agents validate-name\n\n# Output (stdout)\n{\n  \"status\": \"valid\",\n  \"reason\": \"Name meets all quirky validation rules\",\n  \"confidence\": 0.95\n}\n---\n# (stderr - empty if no errors)\n\`\`\`\n\n### Example 2: Validate Haiku via CLI\n\n\`\`\`bash\n# Input\necho \'{\n  \"lines\": [\n    \"Orange sunset fur\",\n    \"Paws dance across the tatami\",\n    \"Zen master purrs\"\n  ]\n}\' | vat agents validate-haiku\n\n# Output (stdout)\n{\n  \"status\": \"valid\",\n  \"syllableCounts\": [5, 7, 5],\n  \"hasKigo\": true,\n  \"kigo\": \"sunset\",\n  \"hasKireji\": false,\n  \"errors\": []\n}\n---\n# (stderr - empty if no errors)\n\`\`\`\n\n### Example 3: Error Handling\n\n\`\`\`bash\n# Invalid input\necho \'{\"invalid\": \"schema\"}\' | vat agents validate-name\n\n# Output\n# (stdout - empty)\n---\nError: Invalid input schema. Expected \'name\' and \'characteristics\' fields.\nSchema validation failed:\n  - Missing required field: name\n  - Missing required field: characteristics\n# (exit code: 2)\n\`\`\`",
-    text: "## CLI Integration Examples\n\n### Example 1: Validate Name via CLI\n\n\`\`\`bash\n# Input\necho \'{\n  \"name\": \"Mr. Whiskers\",\n  \"characteristics\": {\n    \"physical\": {\n      \"furColor\": \"Orange\",\n      \"furPattern\": \"Tabby\",\n      \"size\": \"medium\"\n    },\n    \"behavioral\": {\n      \"personality\": [\"Playful\", \"Curious\"]\n    }\n  }\n}\' | vat agents validate-name\n\n# Output (stdout)\n{\n  \"status\": \"valid\",\n  \"reason\": \"Name meets all quirky validation rules\",\n  \"confidence\": 0.95\n}\n---\n# (stderr - empty if no errors)\n\`\`\`\n\n### Example 2: Validate Haiku via CLI\n\n\`\`\`bash\n# Input\necho \'{\n  \"lines\": [\n    \"Orange sunset fur\",\n    \"Paws dance across the tatami\",\n    \"Zen master purrs\"\n  ]\n}\' | vat agents validate-haiku\n\n# Output (stdout)\n{\n  \"status\": \"valid\",\n  \"syllableCounts\": [5, 7, 5],\n  \"hasKigo\": true,\n  \"kigo\": \"sunset\",\n  \"hasKireji\": false,\n  \"errors\": []\n}\n---\n# (stderr - empty if no errors)\n\`\`\`\n\n### Example 3: Error Handling\n\n\`\`\`bash\n# Invalid input\necho \'{\"invalid\": \"schema\"}\' | vat agents validate-name\n\n# Output\n# (stdout - empty)\n---\nError: Invalid input schema. Expected \'name\' and \'characteristics\' fields.\nSchema validation failed:\n  - Missing required field: name\n  - Missing required field: characteristics\n# (exit code: 2)\n\`\`\`"
-  },
-  nextSteps: {
-    header: "## Next Steps",
-    body: "Once you\'ve successfully orchestrated agents:\n\n1. **Explore Combinations** - Try chaining agents in new ways\n2. **Add Custom Validation** - Create your own quirky validation rules\n3. **Build New Workflows** - Combine agents for complex use cases\n4. **Contribute Patterns** - Share successful orchestration strategies\n5. **Package Skills** - Create distributable skill packages with pure functions",
-    text: "## Next Steps\n\nOnce you\'ve successfully orchestrated agents:\n\n1. **Explore Combinations** - Try chaining agents in new ways\n2. **Add Custom Validation** - Create your own quirky validation rules\n3. **Build New Workflows** - Combine agents for complex use cases\n4. **Contribute Patterns** - Share successful orchestration strategies\n5. **Package Skills** - Create distributable skill packages with pure functions"
-  },
-  questions: {
-    header: "## Questions?",
-    body: "- **For orchestration patterns:** This file (SKILL.md)\n- **For agent-specific details:** See individual agent resources\n- **For implementation examples:** See examples/ directory\n- **For archetype theory:** See package README.md",
-    text: "## Questions?\n\n- **For orchestration patterns:** This file (SKILL.md)\n- **For agent-specific details:** See individual agent resources\n- **For implementation examples:** See examples/ directory\n- **For archetype theory:** See package README.md"
+  advancedPatternsAndCliIntegration: {
+    header: "## Advanced Patterns and CLI Integration",
+    body: "Parallel, conditional and checkpointed orchestration, plus worked CLI request/response pairs,\nare in [orchestration-patterns.md](./orchestration-patterns.md).",
+    text: "## Advanced Patterns and CLI Integration\n\nParallel, conditional and checkpointed orchestration, plus worked CLI request/response pairs,\nare in [orchestration-patterns.md](./orchestration-patterns.md)."
   }
 };

package/dist/generated/resources/skills/orchestration-patterns.d.ts

@@ -0,0 +1,22 @@
+/**
+ * Generated TypeScript declarations - DO NOT EDIT
+ */
+
+export interface Fragment {
+  readonly header: string;
+  readonly body: string;
+  readonly text: string;
+}
+
+export const meta: {};
+
+export const text: string;
+
+export const fragments: {
+  readonly advancedOrchestrationPatterns: Fragment;
+  readonly cliIntegrationExamples: Fragment;
+  readonly nextSteps: Fragment;
+  readonly questions: Fragment;
+};
+
+export type FragmentName = keyof typeof fragments;

package/dist/generated/resources/skills/orchestration-patterns.js

@@ -0,0 +1,30 @@
+/**
+ * Generated from markdown file - DO NOT EDIT
+ */
+
+export const meta = {};
+
+export const text = "# Advanced Orchestration Patterns and CLI Integration\n\nReference material for [SKILL.md](./SKILL.md). Read this when you need the parallel,\nconditional and checkpointed orchestration shapes, or a worked CLI request/response pair.\n\n## Advanced Orchestration Patterns\n\n### Pattern: Parallel Execution\n\nWhen agents don\'t depend on each other, run them in parallel.\n\n**Example:** Generate both name and haiku from same characteristics\n\n\`\`\`typescript\nconst [name, haiku] = await Promise.all([\n  generateCatName(characteristics),\n  generateCatHaiku(characteristics),\n]);\n\`\`\`\n\n**Benefits:** Faster execution, better user experience\n\n### Pattern: Fallback Chain\n\nWhen one agent fails, try alternatives.\n\n**Example:** Photo analysis with text description fallback\n\n\`\`\`typescript\nlet characteristics;\ntry {\n  characteristics = await analyzePhoto(imagePath);\n} catch {\n  // Fallback to text description\n  const description = await getUserDescription();\n  characteristics = await parseDescription(description);\n}\n\`\`\`\n\n**Benefits:** Resilience, better error handling\n\n### Pattern: Conditional Routing\n\nChoose agent path based on user input type.\n\n**Example:** Multi-modal input handling\n\n\`\`\`typescript\nif (input.type === \'image\') {\n  characteristics = await analyzePhoto(input.imagePath);\n} else if (input.type === \'text\') {\n  characteristics = await parseDescription(input.description);\n} else {\n  // Conversational gathering\n  characteristics = await breedAdvisor.gather();\n}\n\`\`\`\n\n**Benefits:** Flexible input handling, better UX\n\n### Pattern: Staged Approval\n\nBreak complex workflows into approval stages.\n\n**Example:** Multi-stage breeding application\n\n\`\`\`typescript\n// Stage 1: Basic info approval\nconst basicApproval = await humanApproval({ stage: \'basic\', data });\nif (!basicApproval.approved) return;\n\n// Stage 2: Detailed questionnaire\nconst detailApproval = await humanApproval({ stage: \'detail\', data });\nif (!detailApproval.approved) return;\n\n// Stage 3: Final review\nconst finalApproval = await humanApproval({ stage: \'final\', data });\n\`\`\`\n\n**Benefits:** Checkpoints prevent wasted work, clearer decision points\n\n## CLI Integration Examples\n\n### Example 1: Validate Name via CLI\n\n\`\`\`bash\n# Input\necho \'{\n  \"name\": \"Mr. Whiskers\",\n  \"characteristics\": {\n    \"physical\": {\n      \"furColor\": \"Orange\",\n      \"furPattern\": \"Tabby\",\n      \"size\": \"medium\"\n    },\n    \"behavioral\": {\n      \"personality\": [\"Playful\", \"Curious\"]\n    }\n  }\n}\' | vat agents validate-name\n\n# Output (stdout)\n{\n  \"status\": \"valid\",\n  \"reason\": \"Name meets all quirky validation rules\",\n  \"confidence\": 0.95\n}\n---\n# (stderr - empty if no errors)\n\`\`\`\n\n### Example 2: Validate Haiku via CLI\n\n\`\`\`bash\n# Input\necho \'{\n  \"lines\": [\n    \"Orange sunset fur\",\n    \"Paws dance across the tatami\",\n    \"Zen master purrs\"\n  ]\n}\' | vat agents validate-haiku\n\n# Output (stdout)\n{\n  \"status\": \"valid\",\n  \"syllableCounts\": [5, 7, 5],\n  \"hasKigo\": true,\n  \"kigo\": \"sunset\",\n  \"hasKireji\": false,\n  \"errors\": []\n}\n---\n# (stderr - empty if no errors)\n\`\`\`\n\n### Example 3: Error Handling\n\n\`\`\`bash\n# Invalid input\necho \'{\"invalid\": \"schema\"}\' | vat agents validate-name\n\n# Output\n# (stdout - empty)\n---\nError: Invalid input schema. Expected \'name\' and \'characteristics\' fields.\nSchema validation failed:\n  - Missing required field: name\n  - Missing required field: characteristics\n# (exit code: 2)\n\`\`\`\n\n## Next Steps\n\nOnce you\'ve successfully orchestrated agents:\n\n1. **Explore Combinations** - Try chaining agents in new ways\n2. **Add Custom Validation** - Create your own quirky validation rules\n3. **Build New Workflows** - Combine agents for complex use cases\n4. **Contribute Patterns** - Share successful orchestration strategies\n5. **Package Skills** - Create distributable skill packages with pure functions\n\n## Questions?\n\n- **For orchestration patterns:** This file (SKILL.md)\n- **For agent-specific details:** See individual agent resources\n- **For implementation examples:** See examples/ directory\n- **For archetype theory:** See package README.md\n";
+
+export const fragments = {
+  advancedOrchestrationPatterns: {
+    header: "## Advanced Orchestration Patterns",
+    body: "### Pattern: Parallel Execution\n\nWhen agents don\'t depend on each other, run them in parallel.\n\n**Example:** Generate both name and haiku from same characteristics\n\n\`\`\`typescript\nconst [name, haiku] = await Promise.all([\n  generateCatName(characteristics),\n  generateCatHaiku(characteristics),\n]);\n\`\`\`\n\n**Benefits:** Faster execution, better user experience\n\n### Pattern: Fallback Chain\n\nWhen one agent fails, try alternatives.\n\n**Example:** Photo analysis with text description fallback\n\n\`\`\`typescript\nlet characteristics;\ntry {\n  characteristics = await analyzePhoto(imagePath);\n} catch {\n  // Fallback to text description\n  const description = await getUserDescription();\n  characteristics = await parseDescription(description);\n}\n\`\`\`\n\n**Benefits:** Resilience, better error handling\n\n### Pattern: Conditional Routing\n\nChoose agent path based on user input type.\n\n**Example:** Multi-modal input handling\n\n\`\`\`typescript\nif (input.type === \'image\') {\n  characteristics = await analyzePhoto(input.imagePath);\n} else if (input.type === \'text\') {\n  characteristics = await parseDescription(input.description);\n} else {\n  // Conversational gathering\n  characteristics = await breedAdvisor.gather();\n}\n\`\`\`\n\n**Benefits:** Flexible input handling, better UX\n\n### Pattern: Staged Approval\n\nBreak complex workflows into approval stages.\n\n**Example:** Multi-stage breeding application\n\n\`\`\`typescript\n// Stage 1: Basic info approval\nconst basicApproval = await humanApproval({ stage: \'basic\', data });\nif (!basicApproval.approved) return;\n\n// Stage 2: Detailed questionnaire\nconst detailApproval = await humanApproval({ stage: \'detail\', data });\nif (!detailApproval.approved) return;\n\n// Stage 3: Final review\nconst finalApproval = await humanApproval({ stage: \'final\', data });\n\`\`\`\n\n**Benefits:** Checkpoints prevent wasted work, clearer decision points",
+    text: "## Advanced Orchestration Patterns\n\n### Pattern: Parallel Execution\n\nWhen agents don\'t depend on each other, run them in parallel.\n\n**Example:** Generate both name and haiku from same characteristics\n\n\`\`\`typescript\nconst [name, haiku] = await Promise.all([\n  generateCatName(characteristics),\n  generateCatHaiku(characteristics),\n]);\n\`\`\`\n\n**Benefits:** Faster execution, better user experience\n\n### Pattern: Fallback Chain\n\nWhen one agent fails, try alternatives.\n\n**Example:** Photo analysis with text description fallback\n\n\`\`\`typescript\nlet characteristics;\ntry {\n  characteristics = await analyzePhoto(imagePath);\n} catch {\n  // Fallback to text description\n  const description = await getUserDescription();\n  characteristics = await parseDescription(description);\n}\n\`\`\`\n\n**Benefits:** Resilience, better error handling\n\n### Pattern: Conditional Routing\n\nChoose agent path based on user input type.\n\n**Example:** Multi-modal input handling\n\n\`\`\`typescript\nif (input.type === \'image\') {\n  characteristics = await analyzePhoto(input.imagePath);\n} else if (input.type === \'text\') {\n  characteristics = await parseDescription(input.description);\n} else {\n  // Conversational gathering\n  characteristics = await breedAdvisor.gather();\n}\n\`\`\`\n\n**Benefits:** Flexible input handling, better UX\n\n### Pattern: Staged Approval\n\nBreak complex workflows into approval stages.\n\n**Example:** Multi-stage breeding application\n\n\`\`\`typescript\n// Stage 1: Basic info approval\nconst basicApproval = await humanApproval({ stage: \'basic\', data });\nif (!basicApproval.approved) return;\n\n// Stage 2: Detailed questionnaire\nconst detailApproval = await humanApproval({ stage: \'detail\', data });\nif (!detailApproval.approved) return;\n\n// Stage 3: Final review\nconst finalApproval = await humanApproval({ stage: \'final\', data });\n\`\`\`\n\n**Benefits:** Checkpoints prevent wasted work, clearer decision points"
+  },
+  cliIntegrationExamples: {
+    header: "## CLI Integration Examples",
+    body: "### Example 1: Validate Name via CLI\n\n\`\`\`bash\n# Input\necho \'{\n  \"name\": \"Mr. Whiskers\",\n  \"characteristics\": {\n    \"physical\": {\n      \"furColor\": \"Orange\",\n      \"furPattern\": \"Tabby\",\n      \"size\": \"medium\"\n    },\n    \"behavioral\": {\n      \"personality\": [\"Playful\", \"Curious\"]\n    }\n  }\n}\' | vat agents validate-name\n\n# Output (stdout)\n{\n  \"status\": \"valid\",\n  \"reason\": \"Name meets all quirky validation rules\",\n  \"confidence\": 0.95\n}\n---\n# (stderr - empty if no errors)\n\`\`\`\n\n### Example 2: Validate Haiku via CLI\n\n\`\`\`bash\n# Input\necho \'{\n  \"lines\": [\n    \"Orange sunset fur\",\n    \"Paws dance across the tatami\",\n    \"Zen master purrs\"\n  ]\n}\' | vat agents validate-haiku\n\n# Output (stdout)\n{\n  \"status\": \"valid\",\n  \"syllableCounts\": [5, 7, 5],\n  \"hasKigo\": true,\n  \"kigo\": \"sunset\",\n  \"hasKireji\": false,\n  \"errors\": []\n}\n---\n# (stderr - empty if no errors)\n\`\`\`\n\n### Example 3: Error Handling\n\n\`\`\`bash\n# Invalid input\necho \'{\"invalid\": \"schema\"}\' | vat agents validate-name\n\n# Output\n# (stdout - empty)\n---\nError: Invalid input schema. Expected \'name\' and \'characteristics\' fields.\nSchema validation failed:\n  - Missing required field: name\n  - Missing required field: characteristics\n# (exit code: 2)\n\`\`\`",
+    text: "## CLI Integration Examples\n\n### Example 1: Validate Name via CLI\n\n\`\`\`bash\n# Input\necho \'{\n  \"name\": \"Mr. Whiskers\",\n  \"characteristics\": {\n    \"physical\": {\n      \"furColor\": \"Orange\",\n      \"furPattern\": \"Tabby\",\n      \"size\": \"medium\"\n    },\n    \"behavioral\": {\n      \"personality\": [\"Playful\", \"Curious\"]\n    }\n  }\n}\' | vat agents validate-name\n\n# Output (stdout)\n{\n  \"status\": \"valid\",\n  \"reason\": \"Name meets all quirky validation rules\",\n  \"confidence\": 0.95\n}\n---\n# (stderr - empty if no errors)\n\`\`\`\n\n### Example 2: Validate Haiku via CLI\n\n\`\`\`bash\n# Input\necho \'{\n  \"lines\": [\n    \"Orange sunset fur\",\n    \"Paws dance across the tatami\",\n    \"Zen master purrs\"\n  ]\n}\' | vat agents validate-haiku\n\n# Output (stdout)\n{\n  \"status\": \"valid\",\n  \"syllableCounts\": [5, 7, 5],\n  \"hasKigo\": true,\n  \"kigo\": \"sunset\",\n  \"hasKireji\": false,\n  \"errors\": []\n}\n---\n# (stderr - empty if no errors)\n\`\`\`\n\n### Example 3: Error Handling\n\n\`\`\`bash\n# Invalid input\necho \'{\"invalid\": \"schema\"}\' | vat agents validate-name\n\n# Output\n# (stdout - empty)\n---\nError: Invalid input schema. Expected \'name\' and \'characteristics\' fields.\nSchema validation failed:\n  - Missing required field: name\n  - Missing required field: characteristics\n# (exit code: 2)\n\`\`\`"
+  },
+  nextSteps: {
+    header: "## Next Steps",
+    body: "Once you\'ve successfully orchestrated agents:\n\n1. **Explore Combinations** - Try chaining agents in new ways\n2. **Add Custom Validation** - Create your own quirky validation rules\n3. **Build New Workflows** - Combine agents for complex use cases\n4. **Contribute Patterns** - Share successful orchestration strategies\n5. **Package Skills** - Create distributable skill packages with pure functions",
+    text: "## Next Steps\n\nOnce you\'ve successfully orchestrated agents:\n\n1. **Explore Combinations** - Try chaining agents in new ways\n2. **Add Custom Validation** - Create your own quirky validation rules\n3. **Build New Workflows** - Combine agents for complex use cases\n4. **Contribute Patterns** - Share successful orchestration strategies\n5. **Package Skills** - Create distributable skill packages with pure functions"
+  },
+  questions: {
+    header: "## Questions?",
+    body: "- **For orchestration patterns:** This file (SKILL.md)\n- **For agent-specific details:** See individual agent resources\n- **For implementation examples:** See examples/ directory\n- **For archetype theory:** See package README.md",
+    text: "## Questions?\n\n- **For orchestration patterns:** This file (SKILL.md)\n- **For agent-specific details:** See individual agent resources\n- **For implementation examples:** See examples/ directory\n- **For archetype theory:** See package README.md"
+  }
+};

package/dist/generated/resources/skills/workflow-orchestration.d.ts

@@ -0,0 +1,19 @@
+/**
+ * Generated TypeScript declarations - DO NOT EDIT
+ */
+
+export interface Fragment {
+  readonly header: string;
+  readonly body: string;
+  readonly text: string;
+}
+
+export const meta: {};
+
+export const text: string;
+
+export const fragments: {
+  readonly detailedWorkflowOrchestration: Fragment;
+};
+
+export type FragmentName = keyof typeof fragments;

package/dist/generated/resources/skills/workflow-orchestration.js

@@ -0,0 +1,15 @@
+/**
+ * Generated from markdown file - DO NOT EDIT
+ */
+
+export const meta = {};
+
+export const text = "# Detailed Workflow Orchestration\n\nReference material for [SKILL.md](./SKILL.md). Read this when you are implementing an\norchestration end to end rather than choosing between the high-level patterns.\n\n## Detailed Workflow Orchestration\n\n### Workflow A: Breed Selection (Conversational)\n\n**When:** User wants help finding the right cat breed.\n\n**Agent:** [breed-advisor](../agents/breed-advisor.md)\n\n**Orchestration strategy:**\n1. **Phase 1 - Gathering** (see [Conversation Strategy](../agents/breed-advisor.md#conversation-strategy))\n   - Collect ≥4 factors including music preference\n   - ONE question at a time (don\'t bombard)\n   - Extract factors after each turn\n   - Monitor readiness: \`factorsCollected >= 4 && musicPreference != null\`\n\n2. **Phase 2 - Recommendation**\n   - Present 3-5 matched breeds\n   - Allow exploration, questions, comparisons\n   - Use conversational formatting (not data dump)\n\n3. **Phase 3 - Selection**\n   - Detect selection signals (\"I\'ll take\", \"sounds good\")\n   - Conclude gracefully\n   - Provide next steps or exit instruction\n\n**Critical factor:** Music preference is the PRIMARY compatibility factor. Ask early, use as conversation anchor. See [Music Preference Insight](../agents/breed-advisor.md#music-preference-insight) for mappings.\n\n**Reference implementation:** See [cat-breed-selection.md](./cat-breed-selection.md) for detailed breed selection orchestration.\n\n### Workflow B: Photo Analysis Pipeline\n\n**When:** User provides a cat photo and wants analysis, name, or haiku.\n\n**Agents:** [photo-analyzer](../agents/photo-analyzer.md) → [name-generator](../agents/name-generator.md) OR [haiku-generator](../agents/haiku-generator.md)\n\n**Orchestration strategy:**\n\n\`\`\`\nStep 1: Analyze Photo\n- Input: Image path/URL\n- Agent: photo-analyzer\n- Output: CatCharacteristics\n- Note: Supports mock mode (reads EXIF) for testing\n\nStep 2: Generate Content\n- Input: CatCharacteristics from Step 1\n- Agent: name-generator OR haiku-generator\n- Output: NameSuggestion OR Haiku\n\nOptional Step 3: Validate\n- Input: Generated content + original characteristics\n- Agent: name-validator OR haiku-validator (pure functions)\n- Output: Validation result\n- If invalid: Retry Step 2 with feedback\n\`\`\`\n\n**Mockable behavior:** photo-analyzer reads EXIF metadata in mock mode. For production, set \`mockable: false\` to use real vision API.\n\n### Workflow C: Text Description Pipeline\n\n**When:** User describes a cat in text (no photo).\n\n**Agents:** [description-parser](../agents/description-parser.md) → [name-generator](../agents/name-generator.md) OR [haiku-generator](../agents/haiku-generator.md)\n\n**Orchestration strategy:**\n\n\`\`\`\nStep 1: Parse Description\n- Input: Text description\n- Agent: description-parser\n- Output: CatCharacteristics (same schema as photo-analyzer!)\n\nStep 2-3: Same as Workflow B\n- Multi-modal convergence: photo and text produce same schema\n- Downstream agents (name-gen, haiku-gen) work with either\n\`\`\`\n\n**Key insight:** Photo and text paths converge at \`CatCharacteristics\` schema. This enables multi-modal workflows without agent changes.\n\n### Workflow D: Generate-Validate-Retry Loop\n\n**When:** Generator produces content that must pass validation rules.\n\n**Pattern:** Generator (LLM) + Validator (pure function) + Retry logic\n\n**Implementation:**\n\n\`\`\`typescript\n// Pseudo-code for orchestration\nlet attempts = 0;\nconst maxAttempts = 3;\n\nwhile (attempts < maxAttempts) {\n  // Generate content\n  const suggestion = await generator(characteristics);\n\n  // Validate (pure function, instant)\n  const validation = validator(suggestion, characteristics);\n\n  if (validation.status === \'valid\') {\n    return suggestion; // Success!\n  }\n\n  // Retry with feedback\n  attempts++;\n  // Optional: Adjust approach based on validation.reason\n}\n\nthrow new Error(\'Could not generate valid content after 3 attempts\');\n\`\`\`\n\n**Why this pattern works:**\n- Generator has NO knowledge of validation rules (isolated concerns)\n- Validator is deterministic (pure function, fast)\n- Feedback loop tests multi-turn orchestration\n- ~60-70% initial rejection rate forces iteration\n\n**Agents using this pattern:**\n- [name-generator](../agents/name-generator.md) + name-validator\n- [haiku-generator](../agents/haiku-generator.md) + haiku-validator\n\n### Workflow E: HITL Approval Gate\n\n**When:** Decision requires human judgment (compliance, taste, ethics).\n\n**Agent:** [human-approval](../agents/human-approval.md)\n\n**Orchestration strategy:**\n\n\`\`\`\nStep 1: Prepare Request\n- Gather all necessary context\n- Format clearly for human reviewer\n- Include relevant data (characteristics, generated content, reasoning)\n\nStep 2: Emit Approval Request\n- Agent: human-approval\n- Input: Request payload\n- Behavior: Blocks waiting for response\n- Timeout: Configurable (default: 60s)\n\nStep 3: Handle Response\n- Approved: Continue workflow\n- Rejected: Handle gracefully (retry, inform user, etc.)\n- Timeout: Fall back to safe default or escalate\n\`\`\`\n\n**Mockable behavior:** Set \`mockable: true\` to auto-approve without human (useful for testing).\n\n**Real-world uses:**\n- Breeding application approval\n- Name selection from alternatives\n- Content moderation decisions\n";
+
+export const fragments = {
+  detailedWorkflowOrchestration: {
+    header: "## Detailed Workflow Orchestration",
+    body: "### Workflow A: Breed Selection (Conversational)\n\n**When:** User wants help finding the right cat breed.\n\n**Agent:** [breed-advisor](../agents/breed-advisor.md)\n\n**Orchestration strategy:**\n1. **Phase 1 - Gathering** (see [Conversation Strategy](../agents/breed-advisor.md#conversation-strategy))\n   - Collect ≥4 factors including music preference\n   - ONE question at a time (don\'t bombard)\n   - Extract factors after each turn\n   - Monitor readiness: \`factorsCollected >= 4 && musicPreference != null\`\n\n2. **Phase 2 - Recommendation**\n   - Present 3-5 matched breeds\n   - Allow exploration, questions, comparisons\n   - Use conversational formatting (not data dump)\n\n3. **Phase 3 - Selection**\n   - Detect selection signals (\"I\'ll take\", \"sounds good\")\n   - Conclude gracefully\n   - Provide next steps or exit instruction\n\n**Critical factor:** Music preference is the PRIMARY compatibility factor. Ask early, use as conversation anchor. See [Music Preference Insight](../agents/breed-advisor.md#music-preference-insight) for mappings.\n\n**Reference implementation:** See [cat-breed-selection.md](./cat-breed-selection.md) for detailed breed selection orchestration.\n\n### Workflow B: Photo Analysis Pipeline\n\n**When:** User provides a cat photo and wants analysis, name, or haiku.\n\n**Agents:** [photo-analyzer](../agents/photo-analyzer.md) → [name-generator](../agents/name-generator.md) OR [haiku-generator](../agents/haiku-generator.md)\n\n**Orchestration strategy:**\n\n\`\`\`\nStep 1: Analyze Photo\n- Input: Image path/URL\n- Agent: photo-analyzer\n- Output: CatCharacteristics\n- Note: Supports mock mode (reads EXIF) for testing\n\nStep 2: Generate Content\n- Input: CatCharacteristics from Step 1\n- Agent: name-generator OR haiku-generator\n- Output: NameSuggestion OR Haiku\n\nOptional Step 3: Validate\n- Input: Generated content + original characteristics\n- Agent: name-validator OR haiku-validator (pure functions)\n- Output: Validation result\n- If invalid: Retry Step 2 with feedback\n\`\`\`\n\n**Mockable behavior:** photo-analyzer reads EXIF metadata in mock mode. For production, set \`mockable: false\` to use real vision API.\n\n### Workflow C: Text Description Pipeline\n\n**When:** User describes a cat in text (no photo).\n\n**Agents:** [description-parser](../agents/description-parser.md) → [name-generator](../agents/name-generator.md) OR [haiku-generator](../agents/haiku-generator.md)\n\n**Orchestration strategy:**\n\n\`\`\`\nStep 1: Parse Description\n- Input: Text description\n- Agent: description-parser\n- Output: CatCharacteristics (same schema as photo-analyzer!)\n\nStep 2-3: Same as Workflow B\n- Multi-modal convergence: photo and text produce same schema\n- Downstream agents (name-gen, haiku-gen) work with either\n\`\`\`\n\n**Key insight:** Photo and text paths converge at \`CatCharacteristics\` schema. This enables multi-modal workflows without agent changes.\n\n### Workflow D: Generate-Validate-Retry Loop\n\n**When:** Generator produces content that must pass validation rules.\n\n**Pattern:** Generator (LLM) + Validator (pure function) + Retry logic\n\n**Implementation:**\n\n\`\`\`typescript\n// Pseudo-code for orchestration\nlet attempts = 0;\nconst maxAttempts = 3;\n\nwhile (attempts < maxAttempts) {\n  // Generate content\n  const suggestion = await generator(characteristics);\n\n  // Validate (pure function, instant)\n  const validation = validator(suggestion, characteristics);\n\n  if (validation.status === \'valid\') {\n    return suggestion; // Success!\n  }\n\n  // Retry with feedback\n  attempts++;\n  // Optional: Adjust approach based on validation.reason\n}\n\nthrow new Error(\'Could not generate valid content after 3 attempts\');\n\`\`\`\n\n**Why this pattern works:**\n- Generator has NO knowledge of validation rules (isolated concerns)\n- Validator is deterministic (pure function, fast)\n- Feedback loop tests multi-turn orchestration\n- ~60-70% initial rejection rate forces iteration\n\n**Agents using this pattern:**\n- [name-generator](../agents/name-generator.md) + name-validator\n- [haiku-generator](../agents/haiku-generator.md) + haiku-validator\n\n### Workflow E: HITL Approval Gate\n\n**When:** Decision requires human judgment (compliance, taste, ethics).\n\n**Agent:** [human-approval](../agents/human-approval.md)\n\n**Orchestration strategy:**\n\n\`\`\`\nStep 1: Prepare Request\n- Gather all necessary context\n- Format clearly for human reviewer\n- Include relevant data (characteristics, generated content, reasoning)\n\nStep 2: Emit Approval Request\n- Agent: human-approval\n- Input: Request payload\n- Behavior: Blocks waiting for response\n- Timeout: Configurable (default: 60s)\n\nStep 3: Handle Response\n- Approved: Continue workflow\n- Rejected: Handle gracefully (retry, inform user, etc.)\n- Timeout: Fall back to safe default or escalate\n\`\`\`\n\n**Mockable behavior:** Set \`mockable: true\` to auto-approve without human (useful for testing).\n\n**Real-world uses:**\n- Breeding application approval\n- Name selection from alternatives\n- Content moderation decisions",
+    text: "## Detailed Workflow Orchestration\n\n### Workflow A: Breed Selection (Conversational)\n\n**When:** User wants help finding the right cat breed.\n\n**Agent:** [breed-advisor](../agents/breed-advisor.md)\n\n**Orchestration strategy:**\n1. **Phase 1 - Gathering** (see [Conversation Strategy](../agents/breed-advisor.md#conversation-strategy))\n   - Collect ≥4 factors including music preference\n   - ONE question at a time (don\'t bombard)\n   - Extract factors after each turn\n   - Monitor readiness: \`factorsCollected >= 4 && musicPreference != null\`\n\n2. **Phase 2 - Recommendation**\n   - Present 3-5 matched breeds\n   - Allow exploration, questions, comparisons\n   - Use conversational formatting (not data dump)\n\n3. **Phase 3 - Selection**\n   - Detect selection signals (\"I\'ll take\", \"sounds good\")\n   - Conclude gracefully\n   - Provide next steps or exit instruction\n\n**Critical factor:** Music preference is the PRIMARY compatibility factor. Ask early, use as conversation anchor. See [Music Preference Insight](../agents/breed-advisor.md#music-preference-insight) for mappings.\n\n**Reference implementation:** See [cat-breed-selection.md](./cat-breed-selection.md) for detailed breed selection orchestration.\n\n### Workflow B: Photo Analysis Pipeline\n\n**When:** User provides a cat photo and wants analysis, name, or haiku.\n\n**Agents:** [photo-analyzer](../agents/photo-analyzer.md) → [name-generator](../agents/name-generator.md) OR [haiku-generator](../agents/haiku-generator.md)\n\n**Orchestration strategy:**\n\n\`\`\`\nStep 1: Analyze Photo\n- Input: Image path/URL\n- Agent: photo-analyzer\n- Output: CatCharacteristics\n- Note: Supports mock mode (reads EXIF) for testing\n\nStep 2: Generate Content\n- Input: CatCharacteristics from Step 1\n- Agent: name-generator OR haiku-generator\n- Output: NameSuggestion OR Haiku\n\nOptional Step 3: Validate\n- Input: Generated content + original characteristics\n- Agent: name-validator OR haiku-validator (pure functions)\n- Output: Validation result\n- If invalid: Retry Step 2 with feedback\n\`\`\`\n\n**Mockable behavior:** photo-analyzer reads EXIF metadata in mock mode. For production, set \`mockable: false\` to use real vision API.\n\n### Workflow C: Text Description Pipeline\n\n**When:** User describes a cat in text (no photo).\n\n**Agents:** [description-parser](../agents/description-parser.md) → [name-generator](../agents/name-generator.md) OR [haiku-generator](../agents/haiku-generator.md)\n\n**Orchestration strategy:**\n\n\`\`\`\nStep 1: Parse Description\n- Input: Text description\n- Agent: description-parser\n- Output: CatCharacteristics (same schema as photo-analyzer!)\n\nStep 2-3: Same as Workflow B\n- Multi-modal convergence: photo and text produce same schema\n- Downstream agents (name-gen, haiku-gen) work with either\n\`\`\`\n\n**Key insight:** Photo and text paths converge at \`CatCharacteristics\` schema. This enables multi-modal workflows without agent changes.\n\n### Workflow D: Generate-Validate-Retry Loop\n\n**When:** Generator produces content that must pass validation rules.\n\n**Pattern:** Generator (LLM) + Validator (pure function) + Retry logic\n\n**Implementation:**\n\n\`\`\`typescript\n// Pseudo-code for orchestration\nlet attempts = 0;\nconst maxAttempts = 3;\n\nwhile (attempts < maxAttempts) {\n  // Generate content\n  const suggestion = await generator(characteristics);\n\n  // Validate (pure function, instant)\n  const validation = validator(suggestion, characteristics);\n\n  if (validation.status === \'valid\') {\n    return suggestion; // Success!\n  }\n\n  // Retry with feedback\n  attempts++;\n  // Optional: Adjust approach based on validation.reason\n}\n\nthrow new Error(\'Could not generate valid content after 3 attempts\');\n\`\`\`\n\n**Why this pattern works:**\n- Generator has NO knowledge of validation rules (isolated concerns)\n- Validator is deterministic (pure function, fast)\n- Feedback loop tests multi-turn orchestration\n- ~60-70% initial rejection rate forces iteration\n\n**Agents using this pattern:**\n- [name-generator](../agents/name-generator.md) + name-validator\n- [haiku-generator](../agents/haiku-generator.md) + haiku-validator\n\n### Workflow E: HITL Approval Gate\n\n**When:** Decision requires human judgment (compliance, taste, ethics).\n\n**Agent:** [human-approval](../agents/human-approval.md)\n\n**Orchestration strategy:**\n\n\`\`\`\nStep 1: Prepare Request\n- Gather all necessary context\n- Format clearly for human reviewer\n- Include relevant data (characteristics, generated content, reasoning)\n\nStep 2: Emit Approval Request\n- Agent: human-approval\n- Input: Request payload\n- Behavior: Blocks waiting for response\n- Timeout: Configurable (default: 60s)\n\nStep 3: Handle Response\n- Approved: Continue workflow\n- Rejected: Handle gracefully (retry, inform user, etc.)\n- Timeout: Fall back to safe default or escalate\n\`\`\`\n\n**Mockable behavior:** Set \`mockable: true\` to auto-approve without human (useful for testing).\n\n**Real-world uses:**\n- Breeding application approval\n- Name selection from alternatives\n- Content moderation decisions"
+  }
+};

package/dist/one-shot-llm-analyzer/description-parser.d.ts

@@ -7,7 +7,7 @@ import { type CatCharacteristics } from '../types/schemas.js';
 export declare const DescriptionParserInputSchema: z.ZodObject<{
     description: z.ZodString;
     mockable: z.ZodOptional<z.ZodBoolean>;
-}, "strip", z.ZodTypeAny, {
+}, "strict", z.ZodTypeAny, {
     description: string;
     mockable?: boolean | undefined;
 }, {

package/dist/one-shot-llm-analyzer/description-parser.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"description-parser.d.ts","sourceRoot":"","sources":["../../src/one-shot-llm-analyzer/description-parser.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,KAAK,EAAE,QAAQ,EAAE,kBAAkB,EAAE,MAAM,4BAA4B,CAAC;AAEtF,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AAKxB,OAAO,EAA4B,KAAK,kBAAkB,EAAE,MAAM,qBAAqB,CAAC;AAGxF;;GAEG;AACH,eAAO,MAAM,4BAA4B;;;;;;;;;EAGvC,CAAC;AAEH,MAAM,MAAM,sBAAsB,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,4BAA4B,CAAC,CAAC;AAElF;;GAEG;AACH,MAAM,WAAW,wBAAwB;IACvC;;;OAGG;IACH,QAAQ,CAAC,EAAE,OAAO,CAAC;CACpB;AAED;;;;;;;;;;;;;;;GAeG;AACH,wBAAsB,gBAAgB,CACpC,WAAW,EAAE,MAAM,EACnB,OAAO,GAAE,wBAA6B,GACrC,OAAO,CAAC,kBAAkB,CAAC,CAQ7B;AAgTD;;;;;;;;GAQG;AACH,eAAO,MAAM,sBAAsB,EAAE,KAAK,CACxC,sBAAsB,EACtB,kBAAkB,CAAC,kBAAkB,EAAE,QAAQ,CAAC,CA0CjD,CAAC"}
+{"version":3,"file":"description-parser.d.ts","sourceRoot":"","sources":["../../src/one-shot-llm-analyzer/description-parser.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,KAAK,EAAE,QAAQ,EAAE,kBAAkB,EAAE,MAAM,4BAA4B,CAAC;AAEtF,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AAKxB,OAAO,EAA4B,KAAK,kBAAkB,EAAE,MAAM,qBAAqB,CAAC;AAGxF;;GAEG;AACH,eAAO,MAAM,4BAA4B;;;;;;;;;EAG9B,CAAC;AAEZ,MAAM,MAAM,sBAAsB,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,4BAA4B,CAAC,CAAC;AAElF;;GAEG;AACH,MAAM,WAAW,wBAAwB;IACvC;;;OAGG;IACH,QAAQ,CAAC,EAAE,OAAO,CAAC;CACpB;AAED;;;;;;;;;;;;;;;GAeG;AACH,wBAAsB,gBAAgB,CACpC,WAAW,EAAE,MAAM,EACnB,OAAO,GAAE,wBAA6B,GACrC,OAAO,CAAC,kBAAkB,CAAC,CAQ7B;AAgTD;;;;;;;;GAQG;AACH,eAAO,MAAM,sBAAsB,EAAE,KAAK,CACxC,sBAAsB,EACtB,kBAAkB,CAAC,kBAAkB,EAAE,QAAQ,CAAC,CA0CjD,CAAC"}

package/dist/one-shot-llm-analyzer/description-parser.js

@@ -9,7 +9,7 @@ import { extractFurColor as extractFurColorUtil } from '../utils/color-extractio
 export const DescriptionParserInputSchema = z.object({
     description: z.string().describe('Text description of the cat (structured or unstructured)'),
     mockable: z.boolean().optional().describe('Whether to use mock mode (default: true)'),
-});
+}).strict();
 /**
  * Parses a text description of a cat and extracts characteristics.
  *

package/dist/one-shot-llm-analyzer/description-parser.js.map

@@ -1 +1 @@
-{"version":3,"file":"description-parser.js","sourceRoot":"","sources":["../../src/one-shot-llm-analyzer/description-parser.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,kBAAkB,EAAE,MAAM,mCAAmC,CAAC;AAEvE,OAAO,EAAE,kBAAkB,EAAE,YAAY,EAAE,MAAM,4BAA4B,CAAC;AAC9E,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AAKxB,OAAO,EAAE,wBAAwB,EAA2B,MAAM,qBAAqB,CAAC;AACxF,OAAO,EAAE,eAAe,IAAI,mBAAmB,EAAE,MAAM,8BAA8B,CAAC;AAEtF;;GAEG;AACH,MAAM,CAAC,MAAM,4BAA4B,GAAG,CAAC,CAAC,MAAM,CAAC;IACnD,WAAW,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,CAAC,0DAA0D,CAAC;IAC5F,QAAQ,EAAE,CAAC,CAAC,OAAO,EAAE,CAAC,QAAQ,EAAE,CAAC,QAAQ,CAAC,0CAA0C,CAAC;CACtF,CAAC,CAAC;AAeH;;;;;;;;;;;;;;;GAeG;AACH,MAAM,CAAC,KAAK,UAAU,gBAAgB,CACpC,WAAmB,EACnB,UAAoC,EAAE;IAEtC,MAAM,EAAE,QAAQ,GAAG,IAAI,EAAE,GAAG,OAAO,CAAC;IAEpC,IAAI,QAAQ,EAAE,CAAC;QACb,OAAO,oBAAoB,CAAC,WAAW,CAAC,CAAC;IAC3C,CAAC;IAED,MAAM,IAAI,KAAK,CAAC,uEAAuE,CAAC,CAAC;AAC3F,CAAC;AAED;;GAEG;AACH,SAAS,oBAAoB,CAAC,WAAmB;IAC/C,MAAM,SAAS,GAAG,WAAW,CAAC,WAAW,EAAE,CAAC;IAE5C,mCAAmC;IACnC,MAAM,QAAQ,GAAG,eAAe,CAAC,SAAS,CAAC,CAAC;IAC5C,MAAM,UAAU,GAAG,iBAAiB,CAAC,SAAS,CAAC,CAAC;IAChD,MAAM,QAAQ,GAAG,eAAe,CAAC,SAAS,CAAC,CAAC;IAC5C,MAAM,KAAK,GAAG,YAAY,CAAC,SAAS,CAAC,CAAC;IACtC,MAAM,IAAI,GAAG,WAAW,CAAC,SAAS,CAAC,CAAC;IAEpC,qCAAqC;IACrC,MAAM,WAAW,GAAG,kBAAkB,CAAC,SAAS,CAAC,CAAC;IAClD,MAAM,MAAM,GAAG,aAAa,CAAC,SAAS,CAAC,CAAC;IACxC,MAAM,aAAa,GAAG,oBAAoB,CAAC,SAAS,CAAC,CAAC;IAEtD,mBAAmB;IACnB,MAAM,GAAG,GAAG,UAAU,CAAC,SAAS,CAAC,CAAC;IAClC,MAAM,MAAM,GAAG,aAAa,CAAC,SAAS,CAAC,CAAC;IACxC,MAAM,UAAU,GAAG,iBAAiB,CAAC,SAAS,CAAC,CAAC;IAEhD,OAAO;QACL,QAAQ,EAAE;YACR,QAAQ;YACR,UAAU;YACV,QAAQ;YACR,KAAK;YACL,IAAI;SACL;QACD,UAAU,EAAE;YACV,WAAW;YACX,MAAM;YACN,aAAa;SACd;QACD,QAAQ,EAAE,GAAG,IAAI,MAAM,IAAI,UAAU,CAAC,CAAC,CAAC;YACtC,GAAG;YACH,MAAM;YACN,UAAU;SACX,CAAC,CAAC,CAAC,SAAS;QACb,WAAW;KACZ,CAAC;AACJ,CAAC;AAED,SAAS,eAAe,CAAC,IAAY;IACnC,OAAO,mBAAmB,CAAC,IAAI,EAAE,cAAc,CAAC,CAAC;AACnD,CAAC;AAED,SAAS,iBAAiB,CAAC,IAAY;IACrC,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,EAAE,CAAC;QACtD,OAAO,OAAO,CAAC;IACjB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC;QAC1B,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,EAAE,CAAC;QAC5B,OAAO,QAAQ,CAAC;IAClB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,SAAS,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,WAAW,CAAC,CAAC,EAAE,CAAC;QACvF,OAAO,YAAY,CAAC;IACtB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,SAAS,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,UAAU,CAAC,EAAE,CAAC;QAC1D,OAAO,SAAS,CAAC;IACnB,CAAC;IAED,OAAO,SAAS,CAAC;AACnB,CAAC;AAED,SAAS,eAAe,CAAC,IAAY;IACnC,IAAI,IAAI,CAAC,QAAQ,CAAC,UAAU,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,UAAU,CAAC,EAAE,CAAC;QAC3D,OAAO,MAAM,CAAC;IAChB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,WAAW,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,WAAW,CAAC,EAAE,CAAC;QAC7D,OAAO,OAAO,CAAC;IACjB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,WAAW,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,UAAU,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,YAAY,CAAC,EAAE,CAAC;QAC3F,OAAO,OAAO,CAAC;IACjB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,YAAY,CAAC,EAAE,CAAC;QAChC,OAAO,QAAQ,CAAC;IAClB,CAAC;IAED,OAAO,SAAS,CAAC;AACnB,CAAC;AAED,SAAS,YAAY,CAAC,IAAY;IAChC,MAAM,MAAM,GAAG;QACb,SAAS,EAAE,SAAS,EAAE,YAAY,EAAE,QAAQ,EAAE,SAAS;QACvD,QAAQ,EAAE,mBAAmB,EAAE,eAAe,EAAE,YAAY;QAC5D,cAAc,EAAE,QAAQ,EAAE,WAAW,EAAE,SAAS,EAAE,kBAAkB;KACrE,CAAC;IAEF,KAAK,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;QAC3B,IAAI,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;YACzB,OAAO,KAAK,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,WAAW,EAAE,GAAG,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;QACvF,CAAC;IACH,CAAC;IAED,OAAO,SAAS,CAAC;AACnB,CAAC;AAED,SAAS,WAAW,CAAC,IAAY;IAC/B,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,EAAE,CAAC;QAChF,OAAO,MAAM,CAAC;IAChB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,EAAE,CAAC;QACtD,OAAO,OAAO,CAAC;IACjB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,SAAS,CAAC,EAAE,CAAC;QAChF,OAAO,aAAa,CAAC;IACvB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;QACnD,OAAO,OAAO,CAAC;IACjB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,EAAE,CAAC;QAC5B,OAAO,QAAQ,CAAC;IAClB,CAAC;IAED,OAAO,SAAS,CAAC;AACnB,CAAC;AAED,SAAS,kBAAkB,CAAC,IAAY;IACtC,MAAM,WAAW,GAAa,EAAE,CAAC;IAEjC,kBAAkB;IAClB,IAAI,IAAI,CAAC,QAAQ,CAAC,SAAS,CAAC,EAAE,CAAC;QAC7B,WAAW,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC;IAC9B,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,UAAU,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,cAAc,CAAC,EAAE,CAAC;QAC/D,WAAW,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC;IAC/B,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,SAAS,CAAC,EAAE,CAAC;QAC7B,WAAW,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC;IAC9B,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,WAAW,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,EAAE,CAAC;QAC1D,WAAW,CAAC,IAAI,CAAC,WAAW,CAAC,CAAC;IAChC,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,UAAU,CAAC,EAAE,CAAC;QACvD,WAAW,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;IAC3B,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,aAAa,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;QAC3D,WAAW,CAAC,IAAI,CAAC,aAAa,CAAC,CAAC;IAClC,CAAC;IAED,yBAAyB;IACzB,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,EAAE,CAAC;QACvD,WAAW,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;IAC7B,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,EAAE,CAAC;QACrD,WAAW,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;IAC3B,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,aAAa,CAAC,EAAE,CAAC;QAC3D,WAAW,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;IAC5B,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,WAAW,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;QACzD,WAAW,CAAC,IAAI,CAAC,WAAW,CAAC,CAAC;IAChC,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;QACnD,WAAW,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;IAC1B,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,aAAa,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,SAAS,CAAC,EAAE,CAAC;QAC7D,WAAW,CAAC,IAAI,CAAC,aAAa,CAAC,CAAC;IAClC,CAAC;IAED,2BAA2B;IAC3B,IAAI,WAAW,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QAC7B,WAAW,CAAC,IAAI,CAAC,YAAY,EAAE,aAAa,CAAC,CAAC;IAChD,CAAC;IAED,OAAO,WAAW,CAAC;AACrB,CAAC;AAED,SAAS,aAAa,CAAC,IAAY;IACjC,MAAM,MAAM,GAAa,EAAE,CAAC;IAE5B,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC;QAC1G,MAAM,CAAC,IAAI,CAAC,0BAA0B,CAAC,CAAC;IAC1C,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;QACnD,MAAM,CAAC,IAAI,CAAC,wBAAwB,CAAC,CAAC;IACxC,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,SAAS,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,gBAAgB,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;QAC1F,MAAM,CAAC,IAAI,CAAC,yBAAyB,CAAC,CAAC;IACzC,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,cAAc,CAAC,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;QACxF,MAAM,CAAC,IAAI,CAAC,6BAA6B,CAAC,CAAC;IAC7C,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,WAAW,CAAC,EAAE,CAAC;QACzD,MAAM,CAAC,IAAI,CAAC,wBAAwB,CAAC,CAAC;IACxC,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;QACrD,MAAM,CAAC,IAAI,CAAC,uBAAuB,CAAC,CAAC;IACvC,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,WAAW,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC;QACxD,MAAM,CAAC,IAAI,CAAC,qBAAqB,CAAC,CAAC;IACrC,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;QACnD,MAAM,CAAC,IAAI,CAAC,yBAAyB,CAAC,CAAC;IACzC,CAAC;IAED,OAAO,MAAM,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,SAAS,CAAC;AAChD,CAAC;AAED,SAAS,oBAAoB,CAAC,IAAY;IACxC,MAAM,aAAa,GAAa,EAAE,CAAC;IAEnC,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;QAClD,aAAa,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;IAC9B,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC;QAC1B,aAAa,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;IAC9B,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;QACrD,aAAa,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;IAC/B,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC;QAC1B,aAAa,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;IAC/B,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC;QACnD,aAAa,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;IAC9B,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;QACtD,aAAa,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAC;IACvC,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,WAAW,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;QACpF,aAAa,CAAC,IAAI,CAAC,YAAY,EAAE,kBAAkB,CAAC,CAAC;IACvD,CAAC;IAED,OAAO,aAAa,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,aAAa,CAAC,CAAC,CAAC,SAAS,CAAC;AAC9D,CAAC;AAED,SAAS,UAAU,CAAC,IAAY;IAC9B,sEAAsE;IACtE,gGAAgG;IAChG,+EAA+E;IAC/E,6EAA6E;IAC7E,gFAAgF;IAChF,4EAA4E;IAC5E,gFAAgF;IAChF,MAAM,QAAQ,GAAG,+BAA+B,CAAC;IACjD,MAAM,QAAQ,GAAG,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IACrC,IAAI,QAAQ,EAAE,CAAC;QACb,OAAO,GAAG,MAAM,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,GAAG,CAAC,YAAY,CAAC;IACnD,CAAC;IAED,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,EAAE,CAAC;QAC5B,OAAO,uBAAuB,CAAC;IACjC,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;QAC3B,OAAO,WAAW,CAAC;IACrB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;QAC3B,OAAO,YAAY,CAAC;IACtB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,SAAS,CAAC,EAAE,CAAC;QACxD,OAAO,oBAAoB,CAAC;IAC9B,CAAC;IAED,OAAO,SAAS,CAAC;AACnB,CAAC;AAED,SAAS,aAAa,CAAC,IAAY;IACjC,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,SAAS,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,SAAS,CAAC,EAAE,CAAC;QACpF,OAAO,gBAAgB,CAAC;IAC1B,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;QAC3B,OAAO,cAAc,CAAC;IACxB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,SAAS,CAAC,EAAE,CAAC;QAC7B,OAAO,cAAc,CAAC;IACxB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;QACtD,OAAO,YAAY,CAAC;IACtB,CAAC;IAED,OAAO,SAAS,CAAC;AACnB,CAAC;AAED,SAAS,iBAAiB,CAAC,IAAY;IACrC,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,cAAc,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,EAAE,CAAC;QACxF,OAAO,qBAAqB,CAAC;IAC/B,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,WAAW,CAAC,EAAE,CAAC;QAC1D,OAAO,YAAY,CAAC;IACtB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC;QAC1B,OAAO,UAAU,CAAC;IACpB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,SAAS,CAAC,EAAE,CAAC;QAC7B,OAAO,aAAa,CAAC;IACvB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,WAAW,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,YAAY,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,EAAE,CAAC;QACzF,OAAO,yBAAyB,CAAC;IACnC,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;QAClD,OAAO,qBAAqB,CAAC;IAC/B,CAAC;IAED,OAAO,SAAS,CAAC;AACnB,CAAC;AAED;;;;;;;;GAQG;AACH,MAAM,CAAC,MAAM,sBAAsB,GAG/B;IACF,IAAI,EAAE,oBAAoB;IAC1B,QAAQ,EAAE;QACR,IAAI,EAAE,oBAAoB;QAC1B,OAAO,EAAE,OAAO;QAChB,WAAW,EAAE,sEAAsE;QACnF,SAAS,EAAE,uBAAuB;QAClC,QAAQ,EAAE;YACR,MAAM,EAAE,iBAAiB;YACzB,WAAW,EAAE,wDAAwD;YACrE,mBAAmB,EAAE,IAAI;YACzB,KAAK,EAAE,kBAAkB;SAC1B;KACF;IACD,OAAO,EAAE,KAAK,EAAE,KAA6B,EAAE,EAAE;QAC/C,iBAAiB;QACjB,MAAM,MAAM,GAAG,4BAA4B,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC;QAC7D,IAAI,CAAC,MAAM,CAAC,OAAO,EAAE,CAAC;YACpB,OAAO;gBACL,MAAM,EAAE,EAAE,MAAM,EAAE,YAAY,EAAE,KAAK,EAAE,kBAAkB,EAAE;aAC5D,CAAC;QACJ,CAAC;QAED,MAAM,EAAE,WAAW,EAAE,QAAQ,GAAG,IAAI,EAAE,GAAG,MAAM,CAAC,IAAI,CAAC;QAErD,OAAO,kBAAkB,CAAC;YACxB,QAAQ;YACR,MAAM,EAAE,GAAG,EAAE,CAAC,oBAAoB,CAAC,WAAW,CAAC;YAC/C,MAAM,EAAE,KAAK,IAAI,EAAE;gBACjB,MAAM,IAAI,KAAK,CAAC,uEAAuE,CAAC,CAAC;gBACzF,6CAA6C;gBAC7C,wEAAwE;gBACxE,8EAA8E;YAChF,CAAC;YACD,WAAW,EAAE,CAAC,GAAG,EAAE,EAAE;gBACnB,MAAM,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,GAAa,CAAC,CAAC;gBACzC,OAAO,wBAAwB,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC;YAChD,CAAC;YACD,YAAY,EAAE,qBAAqB;SACpC,CAAC,CAAC;IACL,CAAC;CACF,CAAC"}
+{"version":3,"file":"description-parser.js","sourceRoot":"","sources":["../../src/one-shot-llm-analyzer/description-parser.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,kBAAkB,EAAE,MAAM,mCAAmC,CAAC;AAEvE,OAAO,EAAE,kBAAkB,EAAE,YAAY,EAAE,MAAM,4BAA4B,CAAC;AAC9E,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AAKxB,OAAO,EAAE,wBAAwB,EAA2B,MAAM,qBAAqB,CAAC;AACxF,OAAO,EAAE,eAAe,IAAI,mBAAmB,EAAE,MAAM,8BAA8B,CAAC;AAEtF;;GAEG;AACH,MAAM,CAAC,MAAM,4BAA4B,GAAG,CAAC,CAAC,MAAM,CAAC;IACnD,WAAW,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,CAAC,0DAA0D,CAAC;IAC5F,QAAQ,EAAE,CAAC,CAAC,OAAO,EAAE,CAAC,QAAQ,EAAE,CAAC,QAAQ,CAAC,0CAA0C,CAAC;CACtF,CAAC,CAAC,MAAM,EAAE,CAAC;AAeZ;;;;;;;;;;;;;;;GAeG;AACH,MAAM,CAAC,KAAK,UAAU,gBAAgB,CACpC,WAAmB,EACnB,UAAoC,EAAE;IAEtC,MAAM,EAAE,QAAQ,GAAG,IAAI,EAAE,GAAG,OAAO,CAAC;IAEpC,IAAI,QAAQ,EAAE,CAAC;QACb,OAAO,oBAAoB,CAAC,WAAW,CAAC,CAAC;IAC3C,CAAC;IAED,MAAM,IAAI,KAAK,CAAC,uEAAuE,CAAC,CAAC;AAC3F,CAAC;AAED;;GAEG;AACH,SAAS,oBAAoB,CAAC,WAAmB;IAC/C,MAAM,SAAS,GAAG,WAAW,CAAC,WAAW,EAAE,CAAC;IAE5C,mCAAmC;IACnC,MAAM,QAAQ,GAAG,eAAe,CAAC,SAAS,CAAC,CAAC;IAC5C,MAAM,UAAU,GAAG,iBAAiB,CAAC,SAAS,CAAC,CAAC;IAChD,MAAM,QAAQ,GAAG,eAAe,CAAC,SAAS,CAAC,CAAC;IAC5C,MAAM,KAAK,GAAG,YAAY,CAAC,SAAS,CAAC,CAAC;IACtC,MAAM,IAAI,GAAG,WAAW,CAAC,SAAS,CAAC,CAAC;IAEpC,qCAAqC;IACrC,MAAM,WAAW,GAAG,kBAAkB,CAAC,SAAS,CAAC,CAAC;IAClD,MAAM,MAAM,GAAG,aAAa,CAAC,SAAS,CAAC,CAAC;IACxC,MAAM,aAAa,GAAG,oBAAoB,CAAC,SAAS,CAAC,CAAC;IAEtD,mBAAmB;IACnB,MAAM,GAAG,GAAG,UAAU,CAAC,SAAS,CAAC,CAAC;IAClC,MAAM,MAAM,GAAG,aAAa,CAAC,SAAS,CAAC,CAAC;IACxC,MAAM,UAAU,GAAG,iBAAiB,CAAC,SAAS,CAAC,CAAC;IAEhD,OAAO;QACL,QAAQ,EAAE;YACR,QAAQ;YACR,UAAU;YACV,QAAQ;YACR,KAAK;YACL,IAAI;SACL;QACD,UAAU,EAAE;YACV,WAAW;YACX,MAAM;YACN,aAAa;SACd;QACD,QAAQ,EAAE,GAAG,IAAI,MAAM,IAAI,UAAU,CAAC,CAAC,CAAC;YACtC,GAAG;YACH,MAAM;YACN,UAAU;SACX,CAAC,CAAC,CAAC,SAAS;QACb,WAAW;KACZ,CAAC;AACJ,CAAC;AAED,SAAS,eAAe,CAAC,IAAY;IACnC,OAAO,mBAAmB,CAAC,IAAI,EAAE,cAAc,CAAC,CAAC;AACnD,CAAC;AAED,SAAS,iBAAiB,CAAC,IAAY;IACrC,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,EAAE,CAAC;QACtD,OAAO,OAAO,CAAC;IACjB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC;QAC1B,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,EAAE,CAAC;QAC5B,OAAO,QAAQ,CAAC;IAClB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,SAAS,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,WAAW,CAAC,CAAC,EAAE,CAAC;QACvF,OAAO,YAAY,CAAC;IACtB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,SAAS,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,UAAU,CAAC,EAAE,CAAC;QAC1D,OAAO,SAAS,CAAC;IACnB,CAAC;IAED,OAAO,SAAS,CAAC;AACnB,CAAC;AAED,SAAS,eAAe,CAAC,IAAY;IACnC,IAAI,IAAI,CAAC,QAAQ,CAAC,UAAU,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,UAAU,CAAC,EAAE,CAAC;QAC3D,OAAO,MAAM,CAAC;IAChB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,WAAW,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,WAAW,CAAC,EAAE,CAAC;QAC7D,OAAO,OAAO,CAAC;IACjB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,WAAW,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,UAAU,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,YAAY,CAAC,EAAE,CAAC;QAC3F,OAAO,OAAO,CAAC;IACjB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,YAAY,CAAC,EAAE,CAAC;QAChC,OAAO,QAAQ,CAAC;IAClB,CAAC;IAED,OAAO,SAAS,CAAC;AACnB,CAAC;AAED,SAAS,YAAY,CAAC,IAAY;IAChC,MAAM,MAAM,GAAG;QACb,SAAS,EAAE,SAAS,EAAE,YAAY,EAAE,QAAQ,EAAE,SAAS;QACvD,QAAQ,EAAE,mBAAmB,EAAE,eAAe,EAAE,YAAY;QAC5D,cAAc,EAAE,QAAQ,EAAE,WAAW,EAAE,SAAS,EAAE,kBAAkB;KACrE,CAAC;IAEF,KAAK,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;QAC3B,IAAI,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;YACzB,OAAO,KAAK,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,WAAW,EAAE,GAAG,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;QACvF,CAAC;IACH,CAAC;IAED,OAAO,SAAS,CAAC;AACnB,CAAC;AAED,SAAS,WAAW,CAAC,IAAY;IAC/B,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,EAAE,CAAC;QAChF,OAAO,MAAM,CAAC;IAChB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,EAAE,CAAC;QACtD,OAAO,OAAO,CAAC;IACjB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,SAAS,CAAC,EAAE,CAAC;QAChF,OAAO,aAAa,CAAC;IACvB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;QACnD,OAAO,OAAO,CAAC;IACjB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,EAAE,CAAC;QAC5B,OAAO,QAAQ,CAAC;IAClB,CAAC;IAED,OAAO,SAAS,CAAC;AACnB,CAAC;AAED,SAAS,kBAAkB,CAAC,IAAY;IACtC,MAAM,WAAW,GAAa,EAAE,CAAC;IAEjC,kBAAkB;IAClB,IAAI,IAAI,CAAC,QAAQ,CAAC,SAAS,CAAC,EAAE,CAAC;QAC7B,WAAW,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC;IAC9B,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,UAAU,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,cAAc,CAAC,EAAE,CAAC;QAC/D,WAAW,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC;IAC/B,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,SAAS,CAAC,EAAE,CAAC;QAC7B,WAAW,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC;IAC9B,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,WAAW,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,EAAE,CAAC;QAC1D,WAAW,CAAC,IAAI,CAAC,WAAW,CAAC,CAAC;IAChC,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,UAAU,CAAC,EAAE,CAAC;QACvD,WAAW,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;IAC3B,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,aAAa,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;QAC3D,WAAW,CAAC,IAAI,CAAC,aAAa,CAAC,CAAC;IAClC,CAAC;IAED,yBAAyB;IACzB,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,EAAE,CAAC;QACvD,WAAW,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;IAC7B,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,EAAE,CAAC;QACrD,WAAW,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;IAC3B,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,aAAa,CAAC,EAAE,CAAC;QAC3D,WAAW,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;IAC5B,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,WAAW,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;QACzD,WAAW,CAAC,IAAI,CAAC,WAAW,CAAC,CAAC;IAChC,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;QACnD,WAAW,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;IAC1B,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,aAAa,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,SAAS,CAAC,EAAE,CAAC;QAC7D,WAAW,CAAC,IAAI,CAAC,aAAa,CAAC,CAAC;IAClC,CAAC;IAED,2BAA2B;IAC3B,IAAI,WAAW,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QAC7B,WAAW,CAAC,IAAI,CAAC,YAAY,EAAE,aAAa,CAAC,CAAC;IAChD,CAAC;IAED,OAAO,WAAW,CAAC;AACrB,CAAC;AAED,SAAS,aAAa,CAAC,IAAY;IACjC,MAAM,MAAM,GAAa,EAAE,CAAC;IAE5B,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC;QAC1G,MAAM,CAAC,IAAI,CAAC,0BAA0B,CAAC,CAAC;IAC1C,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;QACnD,MAAM,CAAC,IAAI,CAAC,wBAAwB,CAAC,CAAC;IACxC,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,SAAS,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,gBAAgB,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;QAC1F,MAAM,CAAC,IAAI,CAAC,yBAAyB,CAAC,CAAC;IACzC,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,cAAc,CAAC,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;QACxF,MAAM,CAAC,IAAI,CAAC,6BAA6B,CAAC,CAAC;IAC7C,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,WAAW,CAAC,EAAE,CAAC;QACzD,MAAM,CAAC,IAAI,CAAC,wBAAwB,CAAC,CAAC;IACxC,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;QACrD,MAAM,CAAC,IAAI,CAAC,uBAAuB,CAAC,CAAC;IACvC,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,WAAW,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC;QACxD,MAAM,CAAC,IAAI,CAAC,qBAAqB,CAAC,CAAC;IACrC,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;QACnD,MAAM,CAAC,IAAI,CAAC,yBAAyB,CAAC,CAAC;IACzC,CAAC;IAED,OAAO,MAAM,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,SAAS,CAAC;AAChD,CAAC;AAED,SAAS,oBAAoB,CAAC,IAAY;IACxC,MAAM,aAAa,GAAa,EAAE,CAAC;IAEnC,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;QAClD,aAAa,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;IAC9B,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC;QAC1B,aAAa,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;IAC9B,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;QACrD,aAAa,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;IAC/B,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC;QAC1B,aAAa,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;IAC/B,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC;QACnD,aAAa,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;IAC9B,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;QACtD,aAAa,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAC;IACvC,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,WAAW,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;QACpF,aAAa,CAAC,IAAI,CAAC,YAAY,EAAE,kBAAkB,CAAC,CAAC;IACvD,CAAC;IAED,OAAO,aAAa,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,aAAa,CAAC,CAAC,CAAC,SAAS,CAAC;AAC9D,CAAC;AAED,SAAS,UAAU,CAAC,IAAY;IAC9B,sEAAsE;IACtE,gGAAgG;IAChG,+EAA+E;IAC/E,6EAA6E;IAC7E,gFAAgF;IAChF,4EAA4E;IAC5E,gFAAgF;IAChF,MAAM,QAAQ,GAAG,+BAA+B,CAAC;IACjD,MAAM,QAAQ,GAAG,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IACrC,IAAI,QAAQ,EAAE,CAAC;QACb,OAAO,GAAG,MAAM,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,GAAG,CAAC,YAAY,CAAC;IACnD,CAAC;IAED,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,EAAE,CAAC;QAC5B,OAAO,uBAAuB,CAAC;IACjC,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;QAC3B,OAAO,WAAW,CAAC;IACrB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;QAC3B,OAAO,YAAY,CAAC;IACtB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,SAAS,CAAC,EAAE,CAAC;QACxD,OAAO,oBAAoB,CAAC;IAC9B,CAAC;IAED,OAAO,SAAS,CAAC;AACnB,CAAC;AAED,SAAS,aAAa,CAAC,IAAY;IACjC,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,SAAS,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,SAAS,CAAC,EAAE,CAAC;QACpF,OAAO,gBAAgB,CAAC;IAC1B,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;QAC3B,OAAO,cAAc,CAAC;IACxB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,SAAS,CAAC,EAAE,CAAC;QAC7B,OAAO,cAAc,CAAC;IACxB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;QACtD,OAAO,YAAY,CAAC;IACtB,CAAC;IAED,OAAO,SAAS,CAAC;AACnB,CAAC;AAED,SAAS,iBAAiB,CAAC,IAAY;IACrC,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,cAAc,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,EAAE,CAAC;QACxF,OAAO,qBAAqB,CAAC;IAC/B,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,WAAW,CAAC,EAAE,CAAC;QAC1D,OAAO,YAAY,CAAC;IACtB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC;QAC1B,OAAO,UAAU,CAAC;IACpB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,SAAS,CAAC,EAAE,CAAC;QAC7B,OAAO,aAAa,CAAC;IACvB,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,WAAW,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,YAAY,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,EAAE,CAAC;QACzF,OAAO,yBAAyB,CAAC;IACnC,CAAC;IACD,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;QAClD,OAAO,qBAAqB,CAAC;IAC/B,CAAC;IAED,OAAO,SAAS,CAAC;AACnB,CAAC;AAED;;;;;;;;GAQG;AACH,MAAM,CAAC,MAAM,sBAAsB,GAG/B;IACF,IAAI,EAAE,oBAAoB;IAC1B,QAAQ,EAAE;QACR,IAAI,EAAE,oBAAoB;QAC1B,OAAO,EAAE,OAAO;QAChB,WAAW,EAAE,sEAAsE;QACnF,SAAS,EAAE,uBAAuB;QAClC,QAAQ,EAAE;YACR,MAAM,EAAE,iBAAiB;YACzB,WAAW,EAAE,wDAAwD;YACrE,mBAAmB,EAAE,IAAI;YACzB,KAAK,EAAE,kBAAkB;SAC1B;KACF;IACD,OAAO,EAAE,KAAK,EAAE,KAA6B,EAAE,EAAE;QAC/C,iBAAiB;QACjB,MAAM,MAAM,GAAG,4BAA4B,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC;QAC7D,IAAI,CAAC,MAAM,CAAC,OAAO,EAAE,CAAC;YACpB,OAAO;gBACL,MAAM,EAAE,EAAE,MAAM,EAAE,YAAY,EAAE,KAAK,EAAE,kBAAkB,EAAE;aAC5D,CAAC;QACJ,CAAC;QAED,MAAM,EAAE,WAAW,EAAE,QAAQ,GAAG,IAAI,EAAE,GAAG,MAAM,CAAC,IAAI,CAAC;QAErD,OAAO,kBAAkB,CAAC;YACxB,QAAQ;YACR,MAAM,EAAE,GAAG,EAAE,CAAC,oBAAoB,CAAC,WAAW,CAAC;YAC/C,MAAM,EAAE,KAAK,IAAI,EAAE;gBACjB,MAAM,IAAI,KAAK,CAAC,uEAAuE,CAAC,CAAC;gBACzF,6CAA6C;gBAC7C,wEAAwE;gBACxE,8EAA8E;YAChF,CAAC;YACD,WAAW,EAAE,CAAC,GAAG,EAAE,EAAE;gBACnB,MAAM,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,GAAa,CAAC,CAAC;gBACzC,OAAO,wBAAwB,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC;YAChD,CAAC;YACD,YAAY,EAAE,qBAAqB;SACpC,CAAC,CAAC;IACL,CAAC;CACF,CAAC"}

package/dist/one-shot-llm-analyzer/haiku-generator.d.ts

@@ -12,7 +12,7 @@ export declare const HaikuGeneratorInputSchema: z.ZodObject<{
             eyeColor: z.ZodOptional<z.ZodString>;
             breed: z.ZodOptional<z.ZodString>;
             size: z.ZodOptional<z.ZodEnum<["tiny", "small", "medium", "large", "extra-large"]>>;
-        }, "strip", z.ZodTypeAny, {
+        }, "strict", z.ZodTypeAny, {
             furColor: string;
             furPattern?: string | undefined;
             eyeColor?: string | undefined;
@@ -29,7 +29,7 @@ export declare const HaikuGeneratorInputSchema: z.ZodObject<{
             personality: z.ZodArray<z.ZodString, "many">;
             quirks: z.ZodOptional<z.ZodArray<z.ZodString, "many">>;
             vocalizations: z.ZodOptional<z.ZodArray<z.ZodString, "many">>;
-        }, "strip", z.ZodTypeAny, {
+        }, "strict", z.ZodTypeAny, {
             personality: string[];
             quirks?: string[] | undefined;
             vocalizations?: string[] | undefined;
@@ -42,7 +42,7 @@ export declare const HaikuGeneratorInputSchema: z.ZodObject<{
             origin: z.ZodOptional<z.ZodString>;
             age: z.ZodOptional<z.ZodString>;
             occupation: z.ZodOptional<z.ZodString>;
-        }, "strip", z.ZodTypeAny, {
+        }, "strict", z.ZodTypeAny, {
             origin?: string | undefined;
             age?: string | undefined;
             occupation?: string | undefined;
@@ -53,7 +53,7 @@ export declare const HaikuGeneratorInputSchema: z.ZodObject<{
         }>>;
         notes: z.ZodOptional<z.ZodString>;
         description: z.ZodString;
-    }, "strip", z.ZodTypeAny, {
+    }, "strict", z.ZodTypeAny, {
         physical: {
             furColor: string;
             furPattern?: string | undefined;
@@ -95,7 +95,7 @@ export declare const HaikuGeneratorInputSchema: z.ZodObject<{
         notes?: string | undefined;
     }>;
     mockable: z.ZodOptional<z.ZodBoolean>;
-}, "strip", z.ZodTypeAny, {
+}, "strict", z.ZodTypeAny, {
     characteristics: {
         physical: {
             furColor: string;

package/dist/one-shot-llm-analyzer/haiku-generator.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"haiku-generator.d.ts","sourceRoot":"","sources":["../../src/one-shot-llm-analyzer/haiku-generator.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,KAAK,EAAE,QAAQ,EAAE,kBAAkB,EAAE,MAAM,4BAA4B,CAAC;AACtF,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AAKxB,OAAO,EAA4B,KAAK,KAAK,EAAE,MAAM,qBAAqB,CAAC;AAE3E;;GAEG;AACH,eAAO,MAAM,yBAAyB;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;EAGpC,CAAC;AAEH,MAAM,MAAM,mBAAmB,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,yBAAyB,CAAC,CAAC;AAsB5E;;;;;;;;;GASG;AACH,eAAO,MAAM,mBAAmB,EAAE,KAAK,CACrC,mBAAmB,EACnB,kBAAkB,CAAC,KAAK,EAAE,QAAQ,CAAC,CAoCpC,CAAC"}
+{"version":3,"file":"haiku-generator.d.ts","sourceRoot":"","sources":["../../src/one-shot-llm-analyzer/haiku-generator.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,KAAK,EAAE,QAAQ,EAAE,kBAAkB,EAAE,MAAM,4BAA4B,CAAC;AACtF,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AAKxB,OAAO,EAA4B,KAAK,KAAK,EAAE,MAAM,qBAAqB,CAAC;AAE3E;;GAEG;AACH,eAAO,MAAM,yBAAyB;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;EAG3B,CAAC;AAEZ,MAAM,MAAM,mBAAmB,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,yBAAyB,CAAC,CAAC;AAsB5E;;;;;;;;;GASG;AACH,eAAO,MAAM,mBAAmB,EAAE,KAAK,CACrC,mBAAmB,EACnB,kBAAkB,CAAC,KAAK,EAAE,QAAQ,CAAC,CAoCpC,CAAC"}

package/dist/one-shot-llm-analyzer/haiku-generator.js

@@ -7,7 +7,7 @@ import { CatCharacteristicsSchema } from '../types/schemas.js';
 export const HaikuGeneratorInputSchema = z.object({
     characteristics: CatCharacteristicsSchema.describe('Cat characteristics to inspire the haiku'),
     mockable: z.boolean().optional().describe('Whether to use mock mode (default: true)'),
-});
+}).strict();
 /**
  * Mock implementation that generates simple haikus based on characteristics.
  */

package/dist/one-shot-llm-analyzer/haiku-generator.js.map

@@ -1 +1 @@
-{"version":3,"file":"haiku-generator.js","sourceRoot":"","sources":["../../src/one-shot-llm-analyzer/haiku-generator.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,kBAAkB,EAAE,kBAAkB,EAAE,MAAM,mCAAmC,CAAC;AAE3F,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AAKxB,OAAO,EAAE,wBAAwB,EAAc,MAAM,qBAAqB,CAAC;AAE3E;;GAEG;AACH,MAAM,CAAC,MAAM,yBAAyB,GAAG,CAAC,CAAC,MAAM,CAAC;IAChD,eAAe,EAAE,wBAAwB,CAAC,QAAQ,CAAC,0CAA0C,CAAC;IAC9F,QAAQ,EAAE,CAAC,CAAC,OAAO,EAAE,CAAC,QAAQ,EAAE,CAAC,QAAQ,CAAC,0CAA0C,CAAC;CACtF,CAAC,CAAC;AAIH;;GAEG;AACH,SAAS,iBAAiB,CAAC,eAAyD;IAClF,MAAM,EAAE,QAAQ,EAAE,UAAU,EAAE,GAAG,eAAe,CAAC;IACjD,MAAM,KAAK,GAAG,QAAQ,CAAC,QAAQ,CAAC,WAAW,EAAE,CAAC;IAC9C,MAAM,WAAW,GAAG,CAAC,UAAU,CAAC,WAAW,CAAC,CAAC,CAAC,IAAI,YAAY,CAAC,CAAC,WAAW,EAAE,CAAC;IAE9E,wDAAwD;IACxD,OAAO;QACL,KAAK,EAAE,GAAG,eAAe,CAAC,KAAK,CAAC,oBAAoB;QACpD,KAAK,EAAE,GAAG,eAAe,CAAC,WAAW,CAAC,oBAAoB;QAC1D,KAAK,EAAE,0BAA0B;KAClC,CAAC;AACJ,CAAC;AAED,SAAS,eAAe,CAAC,GAAW;IAClC,OAAO,GAAG,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,WAAW,EAAE,GAAG,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;AACpD,CAAC;AAED;;;;;;;;;GASG;AACH,MAAM,CAAC,MAAM,mBAAmB,GAG5B;IACF,IAAI,EAAE,iBAAiB;IACvB,QAAQ,EAAE;QACR,IAAI,EAAE,iBAAiB;QACvB,OAAO,EAAE,OAAO;QAChB,WAAW,EAAE,0EAA0E;QACvF,SAAS,EAAE,uBAAuB;QAClC,QAAQ,EAAE;YACR,MAAM,EAAE,WAAW;YACnB,WAAW,EAAE,wEAAwE;YACrF,SAAS,EAAE,KAAK;YAChB,KAAK,EAAE,kBAAkB;YACzB,WAAW,EAAE,GAAG;SACjB;KACF;IACD,OAAO,EAAE,KAAK,EAAE,KAA0B,EAAE,EAAE;QAC5C,MAAM,gBAAgB,GAAG,kBAAkB,CACzC,KAAK,EACL,yBAAyB,CAC1B,CAAC;QACF,IAAI,QAAQ,IAAI,gBAAgB,EAAE,CAAC;YACjC,OAAO,gBAAgB,CAAC;QAC1B,CAAC;QAED,MAAM,EAAE,eAAe,EAAE,QAAQ,GAAG,IAAI,EAAE,GAAG,gBAAgB,CAAC;QAE9D,OAAO,kBAAkB,CAAC;YACxB,QAAQ;YACR,MAAM,EAAE,GAAG,EAAE,CAAC,iBAAiB,CAAC,eAAe,CAAC;YAChD,qBAAqB,EAAE,yEAAyE;YAChG,0CAA0C;YAC1C,qEAAqE;YACrE,2EAA2E;SAC5E,CAAC,CAAC;IACL,CAAC;CACF,CAAC"}
+{"version":3,"file":"haiku-generator.js","sourceRoot":"","sources":["../../src/one-shot-llm-analyzer/haiku-generator.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,kBAAkB,EAAE,kBAAkB,EAAE,MAAM,mCAAmC,CAAC;AAE3F,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AAKxB,OAAO,EAAE,wBAAwB,EAAc,MAAM,qBAAqB,CAAC;AAE3E;;GAEG;AACH,MAAM,CAAC,MAAM,yBAAyB,GAAG,CAAC,CAAC,MAAM,CAAC;IAChD,eAAe,EAAE,wBAAwB,CAAC,QAAQ,CAAC,0CAA0C,CAAC;IAC9F,QAAQ,EAAE,CAAC,CAAC,OAAO,EAAE,CAAC,QAAQ,EAAE,CAAC,QAAQ,CAAC,0CAA0C,CAAC;CACtF,CAAC,CAAC,MAAM,EAAE,CAAC;AAIZ;;GAEG;AACH,SAAS,iBAAiB,CAAC,eAAyD;IAClF,MAAM,EAAE,QAAQ,EAAE,UAAU,EAAE,GAAG,eAAe,CAAC;IACjD,MAAM,KAAK,GAAG,QAAQ,CAAC,QAAQ,CAAC,WAAW,EAAE,CAAC;IAC9C,MAAM,WAAW,GAAG,CAAC,UAAU,CAAC,WAAW,CAAC,CAAC,CAAC,IAAI,YAAY,CAAC,CAAC,WAAW,EAAE,CAAC;IAE9E,wDAAwD;IACxD,OAAO;QACL,KAAK,EAAE,GAAG,eAAe,CAAC,KAAK,CAAC,oBAAoB;QACpD,KAAK,EAAE,GAAG,eAAe,CAAC,WAAW,CAAC,oBAAoB;QAC1D,KAAK,EAAE,0BAA0B;KAClC,CAAC;AACJ,CAAC;AAED,SAAS,eAAe,CAAC,GAAW;IAClC,OAAO,GAAG,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,WAAW,EAAE,GAAG,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;AACpD,CAAC;AAED;;;;;;;;;GASG;AACH,MAAM,CAAC,MAAM,mBAAmB,GAG5B;IACF,IAAI,EAAE,iBAAiB;IACvB,QAAQ,EAAE;QACR,IAAI,EAAE,iBAAiB;QACvB,OAAO,EAAE,OAAO;QAChB,WAAW,EAAE,0EAA0E;QACvF,SAAS,EAAE,uBAAuB;QAClC,QAAQ,EAAE;YACR,MAAM,EAAE,WAAW;YACnB,WAAW,EAAE,wEAAwE;YACrF,SAAS,EAAE,KAAK;YAChB,KAAK,EAAE,kBAAkB;YACzB,WAAW,EAAE,GAAG;SACjB;KACF;IACD,OAAO,EAAE,KAAK,EAAE,KAA0B,EAAE,EAAE;QAC5C,MAAM,gBAAgB,GAAG,kBAAkB,CACzC,KAAK,EACL,yBAAyB,CAC1B,CAAC;QACF,IAAI,QAAQ,IAAI,gBAAgB,EAAE,CAAC;YACjC,OAAO,gBAAgB,CAAC;QAC1B,CAAC;QAED,MAAM,EAAE,eAAe,EAAE,QAAQ,GAAG,IAAI,EAAE,GAAG,gBAAgB,CAAC;QAE9D,OAAO,kBAAkB,CAAC;YACxB,QAAQ;YACR,MAAM,EAAE,GAAG,EAAE,CAAC,iBAAiB,CAAC,eAAe,CAAC;YAChD,qBAAqB,EAAE,yEAAyE;YAChG,0CAA0C;YAC1C,qEAAqE;YACrE,2EAA2E;SAC5E,CAAC,CAAC;IACL,CAAC;CACF,CAAC"}

package/dist/one-shot-llm-analyzer/name-generator.d.ts

@@ -12,7 +12,7 @@ export declare const NameGeneratorInputSchema: z.ZodObject<{
             eyeColor: z.ZodOptional<z.ZodString>;
             breed: z.ZodOptional<z.ZodString>;
             size: z.ZodOptional<z.ZodEnum<["tiny", "small", "medium", "large", "extra-large"]>>;
-        }, "strip", z.ZodTypeAny, {
+        }, "strict", z.ZodTypeAny, {
             furColor: string;
             furPattern?: string | undefined;
             eyeColor?: string | undefined;
@@ -29,7 +29,7 @@ export declare const NameGeneratorInputSchema: z.ZodObject<{
             personality: z.ZodArray<z.ZodString, "many">;
             quirks: z.ZodOptional<z.ZodArray<z.ZodString, "many">>;
             vocalizations: z.ZodOptional<z.ZodArray<z.ZodString, "many">>;
-        }, "strip", z.ZodTypeAny, {
+        }, "strict", z.ZodTypeAny, {
             personality: string[];
             quirks?: string[] | undefined;
             vocalizations?: string[] | undefined;
@@ -42,7 +42,7 @@ export declare const NameGeneratorInputSchema: z.ZodObject<{
             origin: z.ZodOptional<z.ZodString>;
             age: z.ZodOptional<z.ZodString>;
             occupation: z.ZodOptional<z.ZodString>;
-        }, "strip", z.ZodTypeAny, {
+        }, "strict", z.ZodTypeAny, {
             origin?: string | undefined;
             age?: string | undefined;
             occupation?: string | undefined;
@@ -53,7 +53,7 @@ export declare const NameGeneratorInputSchema: z.ZodObject<{
         }>>;
         notes: z.ZodOptional<z.ZodString>;
         description: z.ZodString;
-    }, "strip", z.ZodTypeAny, {
+    }, "strict", z.ZodTypeAny, {
         physical: {
             furColor: string;
             furPattern?: string | undefined;
@@ -95,7 +95,7 @@ export declare const NameGeneratorInputSchema: z.ZodObject<{
         notes?: string | undefined;
     }>;
     mockable: z.ZodOptional<z.ZodBoolean>;
-}, "strip", z.ZodTypeAny, {
+}, "strict", z.ZodTypeAny, {
     characteristics: {
         physical: {
             furColor: string;

package/dist/one-shot-llm-analyzer/name-generator.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"name-generator.d.ts","sourceRoot":"","sources":["../../src/one-shot-llm-analyzer/name-generator.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,KAAK,EAAE,QAAQ,EAAE,kBAAkB,EAAE,MAAM,4BAA4B,CAAC;AACtF,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AAKxB,OAAO,EAA4B,KAAK,cAAc,EAAE,MAAM,qBAAqB,CAAC;AAEpF;;GAEG;AACH,eAAO,MAAM,wBAAwB;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;EAGnC,CAAC;AAEH,MAAM,MAAM,kBAAkB,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,wBAAwB,CAAC,CAAC;AAoF1E;;;;;;;;;GASG;AACH,eAAO,MAAM,kBAAkB,EAAE,KAAK,CACpC,kBAAkB,EAClB,kBAAkB,CAAC,cAAc,EAAE,QAAQ,CAAC,CAoC7C,CAAC"}
+{"version":3,"file":"name-generator.d.ts","sourceRoot":"","sources":["../../src/one-shot-llm-analyzer/name-generator.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,KAAK,EAAE,QAAQ,EAAE,kBAAkB,EAAE,MAAM,4BAA4B,CAAC;AACtF,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AAKxB,OAAO,EAA4B,KAAK,cAAc,EAAE,MAAM,qBAAqB,CAAC;AAEpF;;GAEG;AACH,eAAO,MAAM,wBAAwB;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;EAG1B,CAAC;AAEZ,MAAM,MAAM,kBAAkB,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,wBAAwB,CAAC,CAAC;AAoF1E;;;;;;;;;GASG;AACH,eAAO,MAAM,kBAAkB,EAAE,KAAK,CACpC,kBAAkB,EAClB,kBAAkB,CAAC,cAAc,EAAE,QAAQ,CAAC,CAoC7C,CAAC"}

package/dist/one-shot-llm-analyzer/name-generator.js

@@ -7,7 +7,7 @@ import { CatCharacteristicsSchema } from '../types/schemas.js';
 export const NameGeneratorInputSchema = z.object({
     characteristics: CatCharacteristicsSchema.describe('Cat characteristics to base name suggestions on'),
     mockable: z.boolean().optional().describe('Whether to use mock mode (default: true)'),
-});
+}).strict();
 /**
  * Mock implementation that generates deterministic names based on characteristics.
  */

package/dist/one-shot-llm-analyzer/name-generator.js.map

@@ -1 +1 @@
-{"version":3,"file":"name-generator.js","sourceRoot":"","sources":["../../src/one-shot-llm-analyzer/name-generator.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,kBAAkB,EAAE,kBAAkB,EAAE,MAAM,mCAAmC,CAAC;AAE3F,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AAKxB,OAAO,EAAE,wBAAwB,EAAuB,MAAM,qBAAqB,CAAC;AAEpF;;GAEG;AACH,MAAM,CAAC,MAAM,wBAAwB,GAAG,CAAC,CAAC,MAAM,CAAC;IAC/C,eAAe,EAAE,wBAAwB,CAAC,QAAQ,CAAC,iDAAiD,CAAC;IACrG,QAAQ,EAAE,CAAC,CAAC,OAAO,EAAE,CAAC,QAAQ,EAAE,CAAC,QAAQ,CAAC,0CAA0C,CAAC;CACtF,CAAC,CAAC;AAIH;;GAEG;AACH,SAAS,gBAAgB,CAAC,eAAyD;IACjF,MAAM,EAAE,QAAQ,EAAE,UAAU,EAAE,GAAG,eAAe,CAAC;IAEjD,oDAAoD;IACpD,MAAM,WAAW,GAAG,cAAc,CAAC,QAAQ,CAAC,QAAQ,CAAC,CAAC;IACtD,MAAM,gBAAgB,GAAG,UAAU,CAAC,WAAW,CAAC,CAAC,CAAC,IAAI,YAAY,CAAC;IACnE,iGAAiG;IACjG,MAAM,OAAO,GAAG,IAAI,CAAC,MAAM,EAAE,GAAG,GAAG,CAAC,CAAC,kBAAkB;IAEvD,IAAI,IAAY,CAAC;IACjB,IAAI,YAAsB,CAAC;IAE3B,IAAI,OAAO,EAAE,CAAC;QACZ,qBAAqB;QACrB,IAAI,GAAG,GAAG,aAAa,EAAE,IAAI,WAAW,EAAE,CAAC;QAC3C,YAAY,GAAG;YACb,GAAG,aAAa,EAAE,IAAI,gBAAgB,EAAE;YACxC,GAAG,WAAW,QAAQ,gBAAgB,EAAE;SACzC,CAAC;IACJ,CAAC;SAAM,CAAC;QACN,wBAAwB;QACxB,IAAI,GAAG,GAAG,WAAW,IAAI,iBAAiB,CAAC,gBAAgB,CAAC,EAAE,CAAC;QAC/D,YAAY,GAAG;YACb,WAAW,CAAC,QAAQ,CAAC,QAAQ,CAAC;YAC9B,iBAAiB,CAAC,gBAAgB,CAAC;SACpC,CAAC;IACJ,CAAC;IAED,MAAM,SAAS,GAAG,kBAAkB,QAAQ,CAAC,QAAQ,CAAC,WAAW,EAAE,YAAY,gBAAgB,CAAC,WAAW,EAAE,2DAA2D,CAAC;IAEzK,OAAO;QACL,IAAI;QACJ,SAAS;QACT,YAAY;KACb,CAAC;AACJ,CAAC;AAED,SAAS,cAAc,CAAC,KAAa;IACnC,MAAM,QAAQ,GAA2B;QACvC,MAAM,EAAE,WAAW;QACnB,KAAK,EAAE,QAAQ;QACf,KAAK,EAAE,UAAU;QACjB,IAAI,EAAE,KAAK;QACX,YAAY,EAAE,UAAU;QACxB,MAAM,EAAE,SAAS;QACjB,KAAK,EAAE,OAAO;KACf,CAAC;IACF,OAAO,QAAQ,CAAC,KAAK,CAAC,IAAI,QAAQ,CAAC;AACrC,CAAC;AAED,SAAS,aAAa;IACpB,MAAM,MAAM,GAAG,CAAC,KAAK,EAAE,MAAM,EAAE,MAAM,EAAE,SAAS,EAAE,OAAO,EAAE,UAAU,EAAE,MAAM,EAAE,MAAM,CAAC,CAAC;IACvF,iGAAiG;IACjG,OAAO,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,MAAM,EAAE,GAAG,MAAM,CAAC,MAAM,CAAC,CAAC,IAAI,KAAK,CAAC;AACpE,CAAC;AAED,SAAS,iBAAiB,CAAC,MAAc;IACvC,MAAM,QAAQ,GAAG,CAAC,SAAS,EAAE,cAAc,EAAE,WAAW,EAAE,MAAM,EAAE,MAAM,CAAC,CAAC;IAC1E,iGAAiG;IACjG,OAAO,QAAQ,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,MAAM,EAAE,GAAG,QAAQ,CAAC,MAAM,CAAC,CAAC,IAAI,SAAS,CAAC;AAC5E,CAAC;AAED,SAAS,WAAW,CAAC,KAAa;IAChC,MAAM,OAAO,GAA2B;QACtC,MAAM,EAAE,SAAS;QACjB,KAAK,EAAE,MAAM;QACb,KAAK,EAAE,aAAa;QACpB,IAAI,EAAE,QAAQ;QACd,KAAK,EAAE,OAAO;KACf,CAAC;IACF,OAAO,OAAO,CAAC,KAAK,CAAC,IAAI,QAAQ,CAAC;AACpC,CAAC;AAED,SAAS,iBAAiB,CAAC,MAAc;IACvC,MAAM,KAAK,GAAG,CAAC,SAAS,EAAE,MAAM,EAAE,UAAU,EAAE,UAAU,EAAE,OAAO,EAAE,eAAe,CAAC,CAAC;IACpF,iGAAiG;IACjG,OAAO,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,MAAM,EAAE,GAAG,KAAK,CAAC,MAAM,CAAC,CAAC,IAAI,SAAS,CAAC;AACtE,CAAC;AAED;;;;;;;;;GASG;AACH,MAAM,CAAC,MAAM,kBAAkB,GAG3B;IACF,IAAI,EAAE,gBAAgB;IACtB,QAAQ,EAAE;QACR,IAAI,EAAE,gBAAgB;QACtB,OAAO,EAAE,OAAO;QAChB,WAAW,EAAE,kEAAkE;QAC/E,SAAS,EAAE,uBAAuB;QAClC,QAAQ,EAAE;YACR,MAAM,EAAE,mCAAmC;YAC3C,WAAW,EAAE,0DAA0D;YACvE,aAAa,EAAE,QAAQ;YACvB,KAAK,EAAE,kBAAkB;YACzB,WAAW,EAAE,GAAG;SACjB;KACF;IACD,OAAO,EAAE,KAAK,EAAE,KAAyB,EAAE,EAAE;QAC3C,MAAM,gBAAgB,GAAG,kBAAkB,CACzC,KAAK,EACL,wBAAwB,CACzB,CAAC;QACF,IAAI,QAAQ,IAAI,gBAAgB,EAAE,CAAC;YACjC,OAAO,gBAAgB,CAAC;QAC1B,CAAC;QAED,MAAM,EAAE,eAAe,EAAE,QAAQ,GAAG,IAAI,EAAE,GAAG,gBAAgB,CAAC;QAE9D,OAAO,kBAAkB,CAAC;YACxB,QAAQ;YACR,MAAM,EAAE,GAAG,EAAE,CAAC,gBAAgB,CAAC,eAAe,CAAC;YAC/C,qBAAqB,EAAE,wEAAwE;YAC/F,0CAA0C;YAC1C,oEAAoE;YACpE,0EAA0E;SAC3E,CAAC,CAAC;IACL,CAAC;CACF,CAAC"}
+{"version":3,"file":"name-generator.js","sourceRoot":"","sources":["../../src/one-shot-llm-analyzer/name-generator.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,kBAAkB,EAAE,kBAAkB,EAAE,MAAM,mCAAmC,CAAC;AAE3F,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AAKxB,OAAO,EAAE,wBAAwB,EAAuB,MAAM,qBAAqB,CAAC;AAEpF;;GAEG;AACH,MAAM,CAAC,MAAM,wBAAwB,GAAG,CAAC,CAAC,MAAM,CAAC;IAC/C,eAAe,EAAE,wBAAwB,CAAC,QAAQ,CAAC,iDAAiD,CAAC;IACrG,QAAQ,EAAE,CAAC,CAAC,OAAO,EAAE,CAAC,QAAQ,EAAE,CAAC,QAAQ,CAAC,0CAA0C,CAAC;CACtF,CAAC,CAAC,MAAM,EAAE,CAAC;AAIZ;;GAEG;AACH,SAAS,gBAAgB,CAAC,eAAyD;IACjF,MAAM,EAAE,QAAQ,EAAE,UAAU,EAAE,GAAG,eAAe,CAAC;IAEjD,oDAAoD;IACpD,MAAM,WAAW,GAAG,cAAc,CAAC,QAAQ,CAAC,QAAQ,CAAC,CAAC;IACtD,MAAM,gBAAgB,GAAG,UAAU,CAAC,WAAW,CAAC,CAAC,CAAC,IAAI,YAAY,CAAC;IACnE,iGAAiG;IACjG,MAAM,OAAO,GAAG,IAAI,CAAC,MAAM,EAAE,GAAG,GAAG,CAAC,CAAC,kBAAkB;IAEvD,IAAI,IAAY,CAAC;IACjB,IAAI,YAAsB,CAAC;IAE3B,IAAI,OAAO,EAAE,CAAC;QACZ,qBAAqB;QACrB,IAAI,GAAG,GAAG,aAAa,EAAE,IAAI,WAAW,EAAE,CAAC;QAC3C,YAAY,GAAG;YACb,GAAG,aAAa,EAAE,IAAI,gBAAgB,EAAE;YACxC,GAAG,WAAW,QAAQ,gBAAgB,EAAE;SACzC,CAAC;IACJ,CAAC;SAAM,CAAC;QACN,wBAAwB;QACxB,IAAI,GAAG,GAAG,WAAW,IAAI,iBAAiB,CAAC,gBAAgB,CAAC,EAAE,CAAC;QAC/D,YAAY,GAAG;YACb,WAAW,CAAC,QAAQ,CAAC,QAAQ,CAAC;YAC9B,iBAAiB,CAAC,gBAAgB,CAAC;SACpC,CAAC;IACJ,CAAC;IAED,MAAM,SAAS,GAAG,kBAAkB,QAAQ,CAAC,QAAQ,CAAC,WAAW,EAAE,YAAY,gBAAgB,CAAC,WAAW,EAAE,2DAA2D,CAAC;IAEzK,OAAO;QACL,IAAI;QACJ,SAAS;QACT,YAAY;KACb,CAAC;AACJ,CAAC;AAED,SAAS,cAAc,CAAC,KAAa;IACnC,MAAM,QAAQ,GAA2B;QACvC,MAAM,EAAE,WAAW;QACnB,KAAK,EAAE,QAAQ;QACf,KAAK,EAAE,UAAU;QACjB,IAAI,EAAE,KAAK;QACX,YAAY,EAAE,UAAU;QACxB,MAAM,EAAE,SAAS;QACjB,KAAK,EAAE,OAAO;KACf,CAAC;IACF,OAAO,QAAQ,CAAC,KAAK,CAAC,IAAI,QAAQ,CAAC;AACrC,CAAC;AAED,SAAS,aAAa;IACpB,MAAM,MAAM,GAAG,CAAC,KAAK,EAAE,MAAM,EAAE,MAAM,EAAE,SAAS,EAAE,OAAO,EAAE,UAAU,EAAE,MAAM,EAAE,MAAM,CAAC,CAAC;IACvF,iGAAiG;IACjG,OAAO,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,MAAM,EAAE,GAAG,MAAM,CAAC,MAAM,CAAC,CAAC,IAAI,KAAK,CAAC;AACpE,CAAC;AAED,SAAS,iBAAiB,CAAC,MAAc;IACvC,MAAM,QAAQ,GAAG,CAAC,SAAS,EAAE,cAAc,EAAE,WAAW,EAAE,MAAM,EAAE,MAAM,CAAC,CAAC;IAC1E,iGAAiG;IACjG,OAAO,QAAQ,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,MAAM,EAAE,GAAG,QAAQ,CAAC,MAAM,CAAC,CAAC,IAAI,SAAS,CAAC;AAC5E,CAAC;AAED,SAAS,WAAW,CAAC,KAAa;IAChC,MAAM,OAAO,GAA2B;QACtC,MAAM,EAAE,SAAS;QACjB,KAAK,EAAE,MAAM;QACb,KAAK,EAAE,aAAa;QACpB,IAAI,EAAE,QAAQ;QACd,KAAK,EAAE,OAAO;KACf,CAAC;IACF,OAAO,OAAO,CAAC,KAAK,CAAC,IAAI,QAAQ,CAAC;AACpC,CAAC;AAED,SAAS,iBAAiB,CAAC,MAAc;IACvC,MAAM,KAAK,GAAG,CAAC,SAAS,EAAE,MAAM,EAAE,UAAU,EAAE,UAAU,EAAE,OAAO,EAAE,eAAe,CAAC,CAAC;IACpF,iGAAiG;IACjG,OAAO,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,MAAM,EAAE,GAAG,KAAK,CAAC,MAAM,CAAC,CAAC,IAAI,SAAS,CAAC;AACtE,CAAC;AAED;;;;;;;;;GASG;AACH,MAAM,CAAC,MAAM,kBAAkB,GAG3B;IACF,IAAI,EAAE,gBAAgB;IACtB,QAAQ,EAAE;QACR,IAAI,EAAE,gBAAgB;QACtB,OAAO,EAAE,OAAO;QAChB,WAAW,EAAE,kEAAkE;QAC/E,SAAS,EAAE,uBAAuB;QAClC,QAAQ,EAAE;YACR,MAAM,EAAE,mCAAmC;YAC3C,WAAW,EAAE,0DAA0D;YACvE,aAAa,EAAE,QAAQ;YACvB,KAAK,EAAE,kBAAkB;YACzB,WAAW,EAAE,GAAG;SACjB;KACF;IACD,OAAO,EAAE,KAAK,EAAE,KAAyB,EAAE,EAAE;QAC3C,MAAM,gBAAgB,GAAG,kBAAkB,CACzC,KAAK,EACL,wBAAwB,CACzB,CAAC;QACF,IAAI,QAAQ,IAAI,gBAAgB,EAAE,CAAC;YACjC,OAAO,gBAAgB,CAAC;QAC1B,CAAC;QAED,MAAM,EAAE,eAAe,EAAE,QAAQ,GAAG,IAAI,EAAE,GAAG,gBAAgB,CAAC;QAE9D,OAAO,kBAAkB,CAAC;YACxB,QAAQ;YACR,MAAM,EAAE,GAAG,EAAE,CAAC,gBAAgB,CAAC,eAAe,CAAC;YAC/C,qBAAqB,EAAE,wEAAwE;YAC/F,0CAA0C;YAC1C,oEAAoE;YACpE,0EAA0E;SAC3E,CAAC,CAAC;IACL,CAAC;CACF,CAAC"}