feat: add MCP resources for config, reviews, and metrics

[dtsong]

Summary

• Exposes 3 MCP resources: config://, review://, metrics://
• Config resource reads .ai-review.yaml from disk
• Review/metrics resources query Supabase with proper error handling
• 17 tests covering all resource handlers

Test plan

• All 17 new tests pass
• Full suite (254 tests) passes
• Lint clean

Closes #16

🤖 Generated with Claude Code

[github-actions]

AI Code Review

Summary

The PR implements MCP resources for exposing configuration, reviews, and metrics data. The implementation is generally well-structured with good test coverage, but has several critical security and error handling issues that need to be addressed.

Confidence: 0.00 (low)
This PR may require human review

Strengths

• Comprehensive test coverage with 17 tests covering all major code paths
• Clean separation of concerns with dedicated functions for each resource type
• Good error handling for missing environment variables
• Proper JSON serialization with default=str for datetime objects
• Well-structured URI parsing and routing logic

Issues Found

Critical (2)

src/pr_review_agent/mcp/resources.py:94 - security

SQL injection vulnerability when parsing pr_number from URI path without validation
Suggestion: Validate that pr_number is a valid integer before passing to int() and the database query

src/pr_review_agent/mcp/resources.py:103 - logic

IndexError will occur when result.data is empty, causing the function to crash instead of returning the intended error message
Suggestion: Check if result.data exists and has elements before accessing index 0

Major (3)

src/pr_review_agent/mcp/resources.py:73 - logic

Missing exception handling for Supabase client operations which could fail due to network issues, authentication problems, or database errors
Suggestion: Wrap Supabase operations in try-except blocks to handle potential exceptions gracefully

src/pr_review_agent/mcp/resources.py:122 - logic

Missing exception handling for Supabase client operations in metrics function, same issue as review function
Suggestion: Wrap Supabase operations in try-except blocks to handle potential exceptions gracefully

src/pr_review_agent/mcp/resources.py:89 - security

No validation or sanitization of owner and repo parameters that are used in database queries, potential for injection attacks
Suggestion: Validate that owner and repo contain only safe characters (alphanumeric, hyphens, underscores) before using in queries

Minor (2)

src/pr_review_agent/mcp/resources.py:64 - performance

Supabase client is created on every request instead of being reused
Suggestion: Consider creating a shared client instance or connection pool for better performance

src/pr_review_agent/mcp/resources.py:146 - logic

Division by zero protection exists but the confidence calculation could still fail with malformed data
Suggestion: Add additional validation to ensure confidence_score values are numeric before calculation

Concerns

• Critical security vulnerabilities around input validation that could lead to SQL injection
• Missing exception handling for database operations could cause crashes
• No rate limiting or authentication for resource access
• Direct database queries without any caching could impact performance

Questions

• Should there be any authentication/authorization checks before allowing access to review and metrics data?
• Are there any rate limiting considerations for the Supabase queries?
• Should the config resource support reading from different locations or just the current directory?
• Would it be beneficial to add caching for frequently accessed data?

---

_{Model: claude-sonnet-4-20250514 | Cost: $0.0346 |
Tokens: 5355 in / 1237 out}