{
  "id": "page:library-meta",
  "_kind": "Page",
  "_file": "wiki/library/meta.md",
  "_cluster": "wiki",
  "attributes": {
    "nodeKind": "Page",
    "title": "Meta Specialization - Process, Skill, and Agent Creation (Library)",
    "displayName": "Meta Specialization - Process, Skill, and Agent Creation (Library)",
    "slug": "library/meta",
    "articlePath": "wiki/library/meta.md",
    "article": "\n# Meta Specialization - Process, Skill, and Agent Creation\n\n**Category**: Meta/System Specialization\n**Focus**: Creating Domains, Specializations, Processes, Skills, and Agents\n**Scope**: Babysitter SDK Orchestration Framework\n\n## Overview\n\nThe Meta Specialization is a self-referential specialization focused on the creation and management of the Babysitter SDK's organizational structures: domains, specializations, processes, skills, and agents. This specialization codifies the methodology for building and extending the SDK's capability library.\n\nThis specialization enables systematic, quality-gated creation of new capabilities within the Babysitter SDK framework, ensuring consistency, completeness, and adherence to established patterns.\n\n## Roles and Responsibilities\n\n### Process Architect\n\n**Primary Focus**: Designing and implementing orchestration processes\n\n**Core Responsibilities**:\n- Design process workflows with quality gates\n- Define task definitions and agent/skill assignments\n- Implement breakpoints for human-in-the-loop approval\n- Create convergent loops for iterative refinement\n- Document process inputs, outputs, and artifacts\n\n**Key Skills**:\n- Babysitter SDK process definition patterns\n- Task definition (defineTask) syntax\n- Quality gate design\n- Breakpoint placement strategy\n- Artifact management\n\n**Deliverables**:\n- Process JS files\n- Process documentation\n- Input/output schemas\n- Task dependency graphs\n\n### Skill Designer\n\n**Primary Focus**: Creating and maintaining skill definitions\n\n**Core Responsibilities**:\n- Define skill capabilities and boundaries\n- Document skill inputs/outputs\n- Specify allowed tools for each skill\n- Create skill documentation (SKILL.md)\n- Integrate skills with processes\n\n**Key Skills**:\n- SKILL.md frontmatter syntax\n- Tool permission management\n- Skill capability scoping\n- MCP server integration\n- Best practices documentation\n\n**Deliverables**:\n- SKILL.md files\n- Skill README.md files\n- Supporting scripts\n- Reference documentation\n\n### Agent Designer\n\n**Primary Focus**: Creating and maintaining agent definitions\n\n**Core Responsibilities**:\n- Define agent roles and expertise areas\n- Create agent prompt templates\n- Document agent interaction patterns\n- Specify target processes for agents\n- Design agent collaboration patterns\n\n**Key Skills**:\n- AGENT.md frontmatter syntax\n- Prompt engineering\n- Role-based agent design\n- Expertise specification\n- Process integration patterns\n\n**Deliverables**:\n- AGENT.md files\n- Agent README.md files\n- Prompt templates\n- Interaction documentation\n\n### Specialization Curator\n\n**Primary Focus**: Managing specializations and domains\n\n**Core Responsibilities**:\n- Research specialization requirements\n- Identify processes, skills, and agents needed\n- Maintain backlogs and references\n- Ensure consistency across specializations\n- Quality assurance for new specializations\n\n**Key Skills**:\n- Domain analysis\n- Reference research\n- Backlog management\n- Quality assessment\n- Documentation standards\n\n**Deliverables**:\n- README.md and references.md\n- processes-backlog.md\n- skills-agents-backlog.md\n- skills-agents-references.md\n\n## Directory Structure\n\n```\nspecializations/\n├── meta/                           # This specialization\n│   ├── README.md                   # This file\n│   ├── references.md               # External references\n│   ├── processes-backlog.md        # Phase 2 output\n│   ├── skills-agents-backlog.md    # Phase 4 output\n│   ├── skills-agents-references.md # Phase 5 output\n│   ├── *.js                        # Process files (Phase 3)\n│   ├── skills/                     # Skills (Phase 6)\n│   │   └── <skill-name>/\n│   │       ├── SKILL.md\n│   │       ├── README.md\n│   │       └── scripts/\n│   └── agents/                     # Agents (Phase 6)\n│       └── <agent-name>/\n│           ├── AGENT.md\n│           └── README.md\n├── domains/\n│   ├── business/\n│   │   └── <specialization>/\n│   ├── science/\n│   │   └── <specialization>/\n│   └── social-sciences-humanities/\n│       └── <specialization>/\n└── <non-domain-specialization>/    # R&D/Engineering specializations\n    ├── README.md\n    ├── references.md\n    ├── processes-backlog.md\n    ├── skills-agents-backlog.md\n    ├── skills-agents-references.md\n    ├── *.js\n    ├── skills/\n    └── agents/\n```\n\n## Creation Phases\n\n### Phase 1: Research, README, and References\n\n**Objective**: Establish foundation for the specialization\n\n**Activities**:\n- Research the specialization domain\n- Identify key roles and responsibilities\n- Document goals, objectives, and use cases\n- Compile reference materials with links\n- Create README.md with comprehensive overview\n- Create references.md with curated links\n\n**Output Files**:\n- `README.md` - Specialization overview\n- `references.md` - Reference materials\n\n**Quality Criteria**:\n- Clear scope definition\n- Comprehensive role descriptions\n- Relevant references with working links\n- Consistent formatting\n\n### Phase 2: Identifying Processes\n\n**Objective**: Map out all processes for the specialization\n\n**Activities**:\n- Identify methodologies and work patterns\n- Define process boundaries\n- Describe process inputs/outputs\n- Link to reference materials\n- Create backlog with TODO items\n\n**Output Files**:\n- `processes-backlog.md` - List of processes to implement\n\n**Backlog Format**:\n```markdown\n- [ ] process-name.js - Description of process. Reference: [link]\n```\n\n**Quality Criteria**:\n- Comprehensive coverage of domain\n- Clear process descriptions\n- Traceable to references\n- Appropriate granularity\n\n### Phase 3: Create Process Files\n\n**Objective**: Implement process orchestration files\n\n**Activities**:\n- Create JS files for each process\n- Define task definitions with defineTask\n- Implement quality gates and breakpoints\n- Add input/output schemas\n- Document process metadata\n\n**Output Files**:\n- `*.js` - Process implementation files\n\n**Process File Structure**:\n```javascript\n/**\n * @process specialization/process-name\n * @description Process description\n * @inputs { param1: type, param2: type }\n * @outputs { result: type, artifacts: array }\n */\n\nimport { defineTask } from '@a5c-ai/babysitter-sdk';\n\nexport async function process(inputs, ctx) {\n  // Phase implementation with ctx.task() calls\n  // Breakpoints with ctx.breakpoint()\n  // Return structured output\n}\n\n// Task definitions\nexport const taskName = defineTask('task-name', (args, taskCtx) => ({\n  kind: 'agent',\n  title: 'Task title',\n  skill: { name: 'skill-name' },\n  agent: {\n    name: 'agent-name',\n    prompt: { /* ... */ },\n    outputSchema: { /* ... */ }\n  },\n  io: {\n    inputJsonPath: `tasks/${taskCtx.effectId}/input.json`,\n    outputJsonPath: `tasks/${taskCtx.effectId}/result.json`\n  }\n}));\n```\n\n**Quality Criteria**:\n- Proper JSDoc metadata\n- Correct SDK patterns\n- Quality gates at decision points\n- Breakpoints for approvals\n- Proper error handling\n\n### Phase 4: Identify Skills and Agents\n\n**Objective**: Map skills and agents needed for processes\n\n**Activities**:\n- Analyze each process for skill/agent needs\n- Identify reusable skills across processes\n- Define agent roles and expertise\n- Create mapping table (process -> skills/agents)\n- Document in backlog\n\n**Output Files**:\n- `skills-agents-backlog.md` - Skills and agents to create\n\n**Backlog Format**:\n```markdown\n## Skills\n\n- [ ] SK-XX-001: skill-name - Description\n\n## Agents\n\n- [ ] AG-XX-001: agent-name - Description\n\n## Process-to-Skill/Agent Mapping\n\n| Process | Skills | Agents |\n|---------|--------|--------|\n| process.js | SK-XX-001 | AG-XX-001 |\n```\n\n**Quality Criteria**:\n- Complete mapping for all processes\n- Appropriate skill/agent granularity\n- No duplicate functionality\n- Clear naming conventions\n\n### Phase 5: Research References\n\n**Objective**: Find existing skills/agents to reuse or reference\n\n**Activities**:\n- Search GitHub for existing Claude skills\n- Search for relevant MCP servers\n- Document community resources\n- Identify reusable components\n- Add to references file\n\n**Output Files**:\n- `skills-agents-references.md` - External references\n\n**Quality Criteria**:\n- Active repositories with recent commits\n- Relevant functionality\n- Compatible licensing\n- Working links\n\n### Phase 6: Create Skills and Agents\n\n**Objective**: Implement skill and agent definitions\n\n**Activities**:\n- Create directories for each skill/agent\n- Write SKILL.md/AGENT.md files\n- Add README documentation\n- Create supporting scripts if needed\n- Mark completed in backlog\n\n**SKILL.md Structure**:\n```markdown\n---\nname: skill-name\ndescription: Skill description\nallowed-tools: Bash Read Write Edit Glob Grep\nmetadata:\n  author: babysitter-sdk\n  version: \"1.0.0\"\n  category: category\n  backlog-id: SK-XX-001\n---\n\n# skill-name\n\nDetailed skill documentation...\n```\n\n**AGENT.md Structure**:\n```markdown\n---\nname: agent-name\ndescription: Agent description\nrole: Role Category\nexpertise:\n  - Expertise area 1\n  - Expertise area 2\n---\n\n# Agent Name Agent\n\n## Overview\n\nAgent description...\n\n## Capabilities\n\n- Capability 1\n- Capability 2\n\n## Target Processes\n\n- process-name\n\n## Prompt Template\n\n```javascript\n{\n  role: 'Role description',\n  expertise: ['area1', 'area2'],\n  task: 'Task description',\n  guidelines: ['guideline1', 'guideline2'],\n  outputFormat: 'format'\n}\n```\n```\n\n**Quality Criteria**:\n- Valid frontmatter\n- Comprehensive capabilities\n- Process integration documented\n- Prompt templates provided\n\n### Phase 7: Integrate into Processes\n\n**Objective**: Wire skills/agents into process files\n\n**Activities**:\n- Update task definitions with skill/agent names\n- Verify skill.name and agent.name fields\n- Test process with integrated skills/agents\n- Update backlog with checkmarks\n\n**Integration Pattern**:\n```javascript\nexport const taskName = defineTask('task-name', (args, taskCtx) => ({\n  kind: 'agent',\n  title: 'Task title',\n  skill: { name: 'skill-name' },      // <-- Add skill reference\n  agent: {\n    name: 'agent-name',               // <-- Add agent reference\n    prompt: { /* ... */ },\n    outputSchema: { /* ... */ }\n  },\n  // ...\n}));\n```\n\n**Quality Criteria**:\n- All skills/agents referenced\n- Valid skill/agent names\n- Process runs successfully\n- Backlog fully checked off\n\n## Best Practices\n\n### Process Design\n\n1. **Quality Gates**: Include scoring tasks for iterative refinement\n2. **Breakpoints**: Add human approval at critical decisions\n3. **Artifacts**: Track all generated files\n4. **Logging**: Use ctx.log for process observability\n5. **Error Handling**: Handle failures gracefully\n\n### Skill Design\n\n1. **Single Responsibility**: Each skill should do one thing well\n2. **Tool Permissions**: Only allow necessary tools\n3. **Documentation**: Provide comprehensive usage examples\n4. **Integration**: Document process integration points\n5. **Constraints**: Clearly state limitations\n\n### Agent Design\n\n1. **Clear Role**: Define precise role and expertise\n2. **Prompt Templates**: Provide reusable prompts\n3. **Output Schemas**: Specify expected output format\n4. **Collaboration**: Document interaction patterns\n5. **Target Processes**: List applicable processes\n\n### Naming Conventions\n\n- **Processes**: `kebab-case.js` (e.g., `domain-creation.js`)\n- **Skills**: `kebab-case` directory (e.g., `skill-validator/`)\n- **Agents**: `kebab-case` directory (e.g., `process-architect/`)\n- **Backlog IDs**: `SK-XX-NNN` for skills, `AG-XX-NNN` for agents\n\n## Relationship to Other Specializations\n\n**All Specializations**: Meta provides the framework for creating any specialization\n\n**Technical Documentation**: Shares documentation patterns and standards\n\n**Software Architecture**: Shares ADR and design patterns for process design\n\n**Product Management**: Shares backlog management patterns\n\n## Success Metrics\n\n**Completeness**:\n- All phases completed for each specialization\n- All backlog items addressed\n- All skills/agents integrated\n\n**Quality**:\n- Processes run without errors\n- Skills/agents properly referenced\n- Documentation comprehensive\n\n**Consistency**:\n- Uniform file structure\n- Consistent naming conventions\n- Standard frontmatter formats\n\n## Continuous Improvement\n\n1. **Review New Patterns**: Adopt successful patterns from other specializations\n2. **Refactor Common Components**: Extract reusable skills/agents\n3. **Update References**: Keep external references current\n4. **Validate Integrations**: Regularly test process executions\n5. **Gather Feedback**: Incorporate user feedback into improvements\n\n---\n\n## See Also\n\n- **backlog.md**: Master backlog with phase instructions\n- **references.md**: External reference materials\n- **Babysitter SDK Documentation**: SDK patterns and API\n",
    "documents": [
      "specialization:meta"
    ]
  },
  "outgoingEdges": [
    {
      "from": "page:library-meta",
      "to": "specialization:meta",
      "kind": "documents"
    }
  ],
  "incomingEdges": [
    {
      "from": "page:index",
      "to": "page:library-meta",
      "kind": "contains_page"
    }
  ]
}