feat: implement repository analysis agent with recommendation API

[naaa760]

fix: #22

• Add a new RepositoryAnalysisAgent that analyzes GitHub repositories to automatically generate personalized Watchflow rule recommendations. The agent uses LangGraph for multi-step analysis including repository structure examination, contributing guidelines parsing, and pattern-based rule generation with confidence scoring.

Key Features:

• Repository Analysis: Examines workflows, contributors, branch protection, and code patterns
• Contributing Guidelines Parsing: Uses LLM to extract requirements from CONTRIBUTING.md files
• Rule Recommendations: Generates valid Watchflow YAML rules with confidence scores and reasoning
• API Integration: Adds POST /api/v1/rules/recommend endpoint with caching and rate limiting
• Agent Factory: Integrated into the existing agent system via get_agent("repository_analysis")

Files Added:

• src/agents/repository_analysis_agent/ - Complete agent implementation
• src/api/recommendations.py - REST API endpoints
• Updated src/agents/factory.py and src/main.py for integration

Acceptance Criteria:

Analyzes repositories and generates valid rule recommendations
Recommendations include confidence scores and clear reasoning
API endpoint accepts repository identifiers and returns structured data
Integrates seamlessly with existing Watchflow architecture
All recommendations are valid Watchflow rule YAML
Comprehensive error handling and validation

[dkargatzis]

Hello @naaa760 - thanks for the PR, it adds really cool functionality for better / smother UX. Could we please address the pre-commit and test issues? Also please give an example of the expected outcome - what are the rules that this endpoint suggests for a high volume repo.

[naaa760]

@dkargatzis
please check!

[dkargatzis]

@naaa760 I guess we still need to instruct the agent / pass that context to the agent?