Tech Research Skill Builder

Automatically research technical topics and generate comprehensive project-level skills with the latest documentation and best practices.

Overview

This skill enables automated creation of project-level skills through web research. It:

1. Conducts comprehensive web research on specified technical topics
2. Gathers library documentation, best practices, and code examples
3. Structures findings into an organized skill format
4. Generates a complete, ready-to-use skill package

Workflow

Step 1: Parse Request and Plan Research

When a user requests skill creation, identify:

• Topic: The library, framework, or technical domain to research
• Scope: What aspects to cover (API docs, patterns, best practices)
• Output location: Where to create the skill (default: .claude/skills)

Step 2: Execute Comprehensive Research

Conduct research across four categories:

1. Library Documentation

Search for:

• Official documentation (latest version)
• API references and method signatures
• Getting started guides
• Migration guides

Example searches:

• [topic] official documentation 2025
• [topic] API reference latest
• [topic] getting started guide

2. Best Practices

Search for:

• Industry standards and conventions
• Production deployment guidelines
• Security best practices
• Performance optimization

Example searches:

• [topic] best practices 2025
• [topic] production deployment
• [topic] industry standards

3. Code Examples

Search for:

• Real-world usage patterns
• Common implementations
• Integration examples
• Sample projects

Example searches:

• [topic] code examples
• [topic] common patterns
• [topic] example project github

4. Architectural Patterns

Search for:

• Design patterns
• Architecture decisions
• Scalability patterns
• Implementation strategies

Example searches:

• [topic] architecture patterns
• [topic] design patterns
• [topic] implementation strategies

For detailed research strategies, see research-workflow.md.

Step 3: Structure Research Data

Organize findings into this format:

{
  "topic": "Topic name",
  "metadata": {
    "name": "topic-name",
    "description": "Comprehensive description with triggers"
  },
  "library_docs": [
    {
      "title": "Doc title",
      "summary": "Overview",
      "url": "Source URL",
      "key_points": ["Point 1", "Point 2"],
      "content": "Detailed content"
    }
  ],
  "best_practices": [
    {
      "category": "Category name",
      "description": "Practice description",
      "guidelines": ["Guideline 1", "Guideline 2"],
      "source": "Source URL"
    }
  ],
  "code_examples": [
    {
      "title": "Example title",
      "description": "What it demonstrates",
      "code": "Code snippet",
      "language": "python",
      "source": "Source URL"
    }
  ],
  "architectural_patterns": [
    {
      "name": "Pattern name",
      "description": "Pattern overview",
      "use_cases": ["Use case 1", "Use case 2"],
      "trade_offs": "Pros and cons",
      "source": "Source URL"
    }
  ]
}

Save this structured data to a temporary JSON file for skill generation.

Step 4: Generate Skill Package

Use the generate_skill.py script to create the skill:

python .claude/skills/tech-research-skill-builder/scripts/generate_skill.py \
  /tmp/research_data.json \
  .claude/skills

This generates:

• SKILL.md: Core skill file with frontmatter and navigation
• references/core-concepts.md: Fundamental concepts and terminology
• references/patterns.md: Implementation patterns and code examples
• references/best-practices.md: Production guidelines and recommendations
• references/api-reference.md: Detailed API documentation

For skill generation guidelines, see skill-generation-guide.md.

Step 5: Validate and Package

After generation:

1. Validate the skill structure:

   bashCopy

   python /root/.claude/skills/skill-creator/scripts/quick_validate.py \
     .claude/skills/[generated-skill-name]
   

2. Package the skill (if validation passes):

   bashCopy

   python /root/.claude/skills/skill-creator/scripts/package_skill.py \
     .claude/skills/[generated-skill-name]
   

3. Report to user: Provide the skill location and .skill file path

Example Usage

Example 1: Library-Specific Skill

User request:

"Research FastAPI and create a skill for it"

Workflow:

1. Parse: Topic = "FastAPI", Scope = comprehensive
2. Research:
   • FastAPI official docs (latest version)
   • Best practices for production deployment
   • Common patterns (authentication, database integration)
   • Architecture examples
3. Structure: Organize into JSON format
4. Generate: Create skill at .claude/skills/fastapi
5. Validate and package: Create fastapi.skill file

Example 2: Architectural Pattern Skill

User request:

"Create a skill for microservices architecture patterns"

Workflow:

1. Parse: Topic = "microservices architecture", Scope = patterns
2. Research:
   • Microservices design patterns
   • Best practices for service communication
   • Code examples (API gateways, service mesh)
   • Architecture decisions (monolith vs microservices)
3. Structure: Organize findings
4. Generate: Create skill at .claude/skills/microservices-architecture
5. Validate and package

Example 3: Domain-Specific Technical Skill

User request:

"Research authentication best practices and build a skill"

Workflow:

1. Parse: Topic = "authentication", Scope = best practices
2. Research:
   • Authentication patterns (OAuth, JWT, sessions)
   • Security best practices
   • Implementation examples
   • Industry standards
3. Structure: Organize by authentication type
4. Generate: Create skill at .claude/skills/authentication
5. Validate and package

Quality Criteria

Generated skills should meet these standards:

• Current information: From 2025 or latest version
• Comprehensive coverage: All major aspects of the topic
• Practical examples: Real-world code and patterns
• Clear organization: Logical structure with navigation
• Valid structure: Passes skill validation
• Proper triggers: Description includes when to use

Research Depth Guidelines

Adjust research depth based on topic complexity:

Quick (20-30 min): Simple libraries, basic patterns

• 3-5 sources per category
• Focus on official docs
• Basic examples

Medium (1-2 hours): Standard frameworks, common patterns

• 10-15 sources per category
• Include community resources
• Multiple examples

Deep (3-4 hours): Complex systems, architectural patterns

• 20+ sources per category
• Comprehensive coverage
• Edge cases and advanced topics

Troubleshooting

Research yields limited results

• Broaden search terms
• Include alternative names for the technology
• Search for related technologies/patterns

Generated skill has gaps

• Conduct targeted follow-up research
• Manually add missing sections
• Update research data and regenerate

Validation fails

• Check SKILL.md frontmatter format
• Ensure description is comprehensive
• Verify all reference files are linked

Advanced Usage

Custom Research Scope

Modify the research categories in scripts/research_and_build_skill.py to focus on specific aspects:

def collect_research_requirements(self) -> Dict[str, List[str]]:
    return {
        "security_practices": [...],  # Custom category
        "performance_optimization": [...],
        # Add or remove categories as needed
    }

Multiple Topic Skills

For skills covering multiple related topics:

1. Research each topic separately
2. Merge research data
3. Organize references by topic
4. Generate unified skill

Skill Updates

To update an existing skill with new research:

1. Conduct fresh research
2. Merge with existing content
3. Regenerate skill
4. Replace old skill with updated version