Para/cleanup phase 4

CHANGELOG.md

@@ -5,6 +5,46 @@ All notable changes to codetect will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [Unreleased]
+
+## [3.0.0] - TBD
+
+### Removed
+- **v1 indexer** - Removed `--v1` flag and ctags-based indexing
+- **ctags dependency** - Symbol indexing now uses ast-grep exclusively (built-in, no external dependency)
+- **mattn SQLite driver stub** - Only modernc and ncruces drivers supported
+- **v1 documentation** - Removed `docs/v1/` directory
+
+### Improved
+- **Consolidated enrichment logic** - DRY refactor of scope lookup (3 methods → 1 shared method)
+- **Standardized error handling** - Consistent patterns across tool handlers
+- **Consolidated migration files** - Merged type and database migrations into single file
+- **Vector search performance** - Replaced O(n²) bubble sort with O(n log n) sort.Slice (50x-380x faster)
+
+### Added
+- **Makefile lint/fmt/tidy targets** - Code quality tooling
+
+## [2.2.0] - 2026-02-07
+
+### Added
+- **Rich context in search results** (Phase 2a)
+  - Parent scope extraction (function/class containing each result)
+  - Scope kind tracking (function, method, class, etc.)
+  - Context enrichment (3-5 lines before/after matches)
+  - Receiver type for methods
+  - `include_context` parameter for search tools
+
+### Improved
+- **AST chunker** - Extracts scope information during indexing
+- **Search results** - Include rich metadata for better LLM understanding
+- **Dependency injection** - Clean, removable enrichment pattern
+
+### Performance
+- **6.5% token reduction** in evaluations
+- **3.2% accuracy improvement** in evaluations
+
+---
+
 ## [2.1.1] - 2026-02-02
 
 ### Fixed

CLAUDE.md

@@ -4,15 +4,15 @@
 
 ## About
 
-codetect is a local MCP server providing fast codebase search, file retrieval, symbol navigation, and semantic search for Claude Code. It combines keyword search (ripgrep), symbol indexing (ctags), and semantic search (Ollama embeddings) to enable natural language code exploration.
+codetect is a local MCP server providing fast codebase search, file retrieval, symbol navigation, and semantic search for Claude Code. It combines keyword search (ripgrep), symbol indexing (ast-grep), and semantic search (Ollama embeddings) to enable natural language code exploration.
 
 ## Tech Stack
 
 - **Go 1.25+** - Primary language
 - **SQLite** - Default embedded database (modernc.org/sqlite)
 - **PostgreSQL + pgvector** - Optional high-performance vector backend
 - **ripgrep** - Fast keyword search
-- **universal-ctags** - Symbol indexing
+- **tree-sitter (via ast-grep)** - Symbol indexing
 - **Ollama** - Local embeddings for semantic search
 - **MCP (Model Context Protocol)** - LLM tool integration
 
@@ -118,16 +118,6 @@ make clean                    # Clean build artifacts
 - `CODETECT_DB_TYPE` - `sqlite` (default) or `postgres`
 - `CODETECT_DB_DSN` - PostgreSQL connection string
 
-**Project Config (`.codetect.yaml`):**
-```yaml
-db:
-  type: postgres
-  dsn: postgres://user:pass@localhost/codetect
-
-embedding:
-  provider: ollama
-  model: nomic-embed-text
-```
 
 ## MCP Tools
 

Makefile

@@ -84,13 +84,8 @@ doctor:
 	@command -v rg >/dev/null 2>&1 || { echo "❌ missing: ripgrep (rg)"; exit 1; }
 	@echo "✓ ripgrep: $$(rg --version | head -1)"
 	@echo ""
-	@echo "=== Optional (for symbol indexing) ==="
-	@if command -v ctags >/dev/null 2>&1 && ctags --version 2>&1 | grep -q "Universal Ctags"; then \
-		echo "✓ ctags: $$(ctags --version | head -1)"; \
-	else \
-		echo "○ ctags: not found (symbol indexing disabled)"; \
-		echo "  Install with: brew install universal-ctags"; \
-	fi
+	@echo "=== Symbol Indexing ==="
+	@echo "✓ ast-grep: built-in (no external dependency required)"
 	@echo ""
 	@echo "=== Embedding Provider ==="
 	@PROVIDER=$${CODETECT_EMBEDDING_PROVIDER:-ollama}; \
@@ -199,3 +194,16 @@ eval-list: build
 # Show latest evaluation report
 eval-report: build
 	@./$(EVAL) report
+
+# Code quality targets
+lint:
+	@command -v golangci-lint >/dev/null 2>&1 || { echo "golangci-lint not installed. Install from: https://golangci-lint.run/usage/install/"; exit 1; }
+	golangci-lint run ./...
+
+fmt:
+	gofmt -s -w .
+	@command -v goimports >/dev/null 2>&1 && goimports -w . || echo "Note: goimports not found, skipping import formatting"
+
+tidy:
+	go mod tidy
+	go mod verify

README.md

@@ -2,9 +2,9 @@
 
 A local MCP server providing fast codebase search, file retrieval, symbol navigation, and semantic search for Claude Code.
 
-## What's New in v2.0.0 🎉
+## What's New in v2.2.0 🎉
 
-codetect v2.0.0 brings **multi-repo support**, **parallel embedding**, and **improved user experience**:
+codetect v2.2.0 brings **rich context enrichment**, **ast-grep symbol indexing**, and **improved search results**:
 
 - ✨ **Dimension-grouped embedding tables** - Multiple repos can use different embedding models without conflicts
 - ⚡ **Parallel embedding** with `-j` flag - 3.3x faster embedding with configurable workers
@@ -13,15 +13,15 @@ codetect v2.0.0 brings **multi-repo support**, **parallel embedding**, and **imp
 - 🛡️ **Config preservation** - Reinstalls no longer overwrite your settings
 - 🐛 **Better error handling** - Improved ripgrep error messages and diagnostics
 
-**Upgrading from v1.x?** v2.0.0 is fully backward compatible. See [Migration Guide](docs/MIGRATION.md) for details.
+**Upgrading from v2.0.0?** v2.2.0 adds rich context and removes ctags dependency. See [CHANGELOG](CHANGELOG.md) for details.
 
 **Full changelog:** [CHANGELOG.md](CHANGELOG.md)
 
 ## Features
 
 - **`search_keyword`** - Fast regex search powered by ripgrep
 - **`get_file`** - File reading with optional line-range slicing
-- **`find_symbol`** - Symbol lookup (functions, types, etc.) via ctags + SQLite
+- **`find_symbol`** - Symbol lookup (functions, types, etc.) via ast-grep + SQLite
 - **`list_defs_in_file`** - List all definitions in a file
 - **`search_semantic`** - Semantic code search via local embeddings (Ollama)
 - **`hybrid_search`** - Combined keyword + semantic search
@@ -37,7 +37,6 @@ cd codetect
 
 The installer will:
 - ✓ Check for required dependencies (Go, ripgrep)
-- ✓ Offer to install ctags automatically for symbol indexing
 - ✓ Guide you through Ollama setup for semantic search (with prominent warnings if missing)
 - ✓ Build and install globally to `~/.local/bin`
 - ✓ Configure your shell PATH automatically
@@ -62,39 +61,32 @@ See [Installation Guide](docs/installation.md) for detailed setup instructions.
 |------------|----------|---------|
 | Go 1.21+ | Yes | Building from source |
 | [ripgrep](https://github.com/BurntSushi/ripgrep) | Yes | Keyword search |
-| [universal-ctags](https://github.com/universal-ctags/ctags) | No | Symbol indexing (v1 legacy mode only, v2 uses built-in tree-sitter) |
 | [Ollama](https://ollama.ai) | No | Semantic search (local embeddings) |
 
-**Note:** v2 (default) uses built-in tree-sitter parsers for symbol extraction. ctags is only needed if using `--v1` legacy mode.
+**Note:** Symbol indexing uses built-in ast-grep (no external dependencies required).
 
 ## CLI Commands
 
 ### Main Commands
 
 ```bash
 codetect init        # Initialize in current directory (.mcp.json)
-codetect index       # Index with v2 (AST-based, incremental, 15x faster)
-codetect index --v1  # Index with v1 (ctags-based, legacy, deprecated)
+codetect index       # Index symbols (AST-based, incremental)
 codetect embed       # Generate embeddings (sequential)
 codetect embed -j 10 # Generate embeddings in parallel (10 workers, 3.3x faster)
 codetect doctor      # Check dependencies and configuration
-codetect stats       # Show v2 index statistics
-codetect stats --v1  # Show v1 index statistics (if v1 index exists)
+codetect stats       # Show index statistics
 codetect migrate     # Discover existing indexes and register them
 codetect update      # Update to latest version
 codetect help        # Show all commands
 ```
 
-**v2 features (default):**
+**Key features:**
 - ⚡ Incremental indexing with Merkle tree change detection (~2s vs ~30s)
 - 🧬 AST-based chunking preserves semantic boundaries
 - 📦 Content-addressed caching (95% cache hit rate)
 - 🔄 Parallel embedding with `-j` flag (3.3x faster)
 
-**v1 legacy mode:**
-- Use `--v1` flag for ctags-based indexing (deprecated, removed in v3.0.0)
-- See [v1 documentation](docs/v1/README.md) for details
-
 ### Daemon Commands
 
 ```bash
@@ -279,7 +271,7 @@ See [MCP Compatibility](docs/mcp-compatibility.md) for details and roadmap for n
 
 - [x] MCP stdio server
 - [x] Keyword search via ripgrep
-- [x] Symbol indexing via ctags
+- [x] Symbol indexing via ast-grep
 - [x] Semantic search via Ollama
 - [x] Hybrid search
 - [x] Global installation