Registry Review MCP Server

# Quick Reference: Top 10 Improvements **Last Updated:** 2025-11-15 **For Full Details:** See `REFACTORING_ACTION_PLAN.md` and `CODE_QUALITY_REVIEW.md` --- ## 🔥 Critical (Do First) ### 1. Methodology Registry **Problem:** "soil-carbon-v1.2.2" hardcoded in 6+ files **Impact:** Cannot support multiple methodologies **Effort:** 4-6 hours **File:** Create `src/registry_review_mcp/config/methodologies.py` ### 2. Document Classifier Registry **Problem:** 80-line if/elif chain for classification **Impact:** Cannot add document types without code changes **Effort:** 6-8 hours **File:** Create `src/registry_review_mcp/classifiers/` --- ## 🔶 High Priority (Next) ### 3. Extractor Factory **Problem:** Each extractor creates own Anthropic client **Impact:** Poor resource management, hard to test **Effort:** 3-4 hours **File:** Create `src/registry_review_mcp/extractors/factory.py` ### 4. Refactor discover_documents (195 lines) **Problem:** Monolithic function with multiple responsibilities **Impact:** Hard to test, understand, modify **Effort:** 3-4 hours **File:** `src/registry_review_mcp/tools/document_tools.py:67-261` ### 5. Refactor cross_validate (216 lines) **Problem:** Longest function in codebase **Impact:** Complex, fragile, untestable **Effort:** 3-4 hours **File:** `src/registry_review_mcp/tools/validation_tools.py:486-701` ### 6. Refactor generate_review_report (208 lines) **Problem:** Report generation mixed with formatting **Impact:** Hard to customize or extend **Effort:** 2-3 hours **File:** `src/registry_review_mcp/tools/report_tools.py:18-225` --- ## 🟡 Medium Priority (Quality) ### 7. Error Handling **Problem:** Silent failures, no logging **Impact:** Hard to debug, poor UX **Effort:** 4-6 hours **Files:** Multiple (A_initialize.py:86, document_tools.py:172, etc.) ### 8. Move Magic Numbers to Config **Problem:** Hardcoded thresholds (0.7, 5%, 9000, etc.) **Impact:** Cannot configure per environment **Effort:** 2-3 hours **File:** `src/registry_review_mcp/config/settings.py` ### 9. Citation Parser **Problem:** String matching for citations **Impact:** Brittle, error-prone **Effort:** 3-4 hours **File:** Create `src/registry_review_mcp/utils/citations.py` ### 10. Validation Registry **Problem:** Validation rules hardcoded **Impact:** Cannot customize per methodology **Effort:** 4-5 hours **File:** Create `src/registry_review_mcp/validators/registry.py` --- ## Quick Stats - **Total Issues Found:** 47 - **Total Effort:** 8-10 days - **Code Reduction:** ~40% duplication removed - **Extensibility Gain:** Infinite (methodology/classifier/validator plugins) --- ## Before & After Examples ### Methodology Usage ```python # Before (HARDCODED) methodology: str = "soil-carbon-v1.2.2" # After (DYNAMIC) from ..config.methodologies import get_methodology_registry registry = get_methodology_registry() methodology: str = registry.get_default() ``` ### Document Classification ```python # Before (IF/ELIF CHAIN) if match_any(filename, PROJECT_PLAN_PATTERNS): return ("project_plan", 0.95, "filename") elif match_any(filename, BASELINE_PATTERNS): return ("baseline_report", 0.90, "filename") # ... 10+ more branches # After (STRATEGY PATTERN) registry = ClassifierRegistry() registry.register(PatternClassifier("project_plan", patterns, 0.95)) doc_type, confidence, method = registry.classify(filepath) ``` ### Long Function ```python # Before (195 LINES) async def discover_documents(session_id: str) -> dict: # validation # file scanning # processing loop # duplicate checking # classification # error handling # saving # formatting # After (5 FOCUSED FUNCTIONS) async def discover_documents(session_id: str) -> dict: session_data = await _validate_session(session_id) files = await _scan_directory(Path(session_data["..."])) results = await _process_files(files, session_id) await _save_results(session_id, results) return _format_response(session_id, results) ``` --- ## Implementation Checklist ### Week 1 - [ ] Methodology Registry (#1) - [ ] Document Classifier Registry (#2) - [ ] Integration tests ### Week 2 - [ ] Extractor Factory (#3) - [ ] Refactor discover_documents (#4) - [ ] Refactor cross_validate (#5) - [ ] Refactor generate_review_report (#6) ### Week 3 (Optional) - [ ] Error handling (#7) - [ ] Config for magic numbers (#8) - [ ] Citation parser (#9) - [ ] Validation registry (#10) --- ## Testing Strategy **For each refactoring:** 1. ✅ Write tests capturing current behavior 2. ✅ Refactor code 3. ✅ Ensure all tests pass 4. ✅ Add tests for new functionality 5. ✅ Run full test suite (184+ tests) --- ## Success Criteria After completing all 10 improvements: ✅ Zero hardcoded "soil-carbon-v1.2.2" strings ✅ No functions >100 lines ✅ 40% reduction in code duplication ✅ All magic numbers in configuration ✅ Can add methodologies without code changes ✅ Can add classifiers without code changes ✅ Comprehensive error logging ✅ All 184+ tests passing --- ## Questions? **Full Details:** See `docs/REFACTORING_ACTION_PLAN.md` (915 lines) **Complete Analysis:** See `docs/CODE_QUALITY_REVIEW.md` (2,252 lines) **Bug Fixes:** See `docs/BUG_FIXES_2025-11-14.md` (337 lines) **Architecture:** See `docs/CODE_REVIEW_INITIALIZATION.md` (884 lines)