codexa 0.4.3__tar.gz → 0.5.0__tar.gz

{codexa-0.4.3 → codexa-0.5.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: codexa
-Version: 0.4.3
+Version: 0.5.0
 Summary: Developer intelligence CLI — semantic code search, AI-assisted understanding, and agent tooling for codebases
 License: MIT
 Project-URL: Homepage, https://codex-a.dev
@@ -65,9 +65,11 @@ Dynamic: license-file
 
 <p align="center">
   <a href="https://github.com/M9nx/CodexA/actions"><img src="https://github.com/M9nx/CodexA/actions/workflows/ci.yml/badge.svg" alt="CI"></a>
+  <a href="https://pypi.org/project/codexa/"><img src="https://img.shields.io/pypi/v/codexa?color=blue&label=PyPI" alt="PyPI"></a>
+  <a href="https://pepy.tech/project/codexa"><img src="https://img.shields.io/pepy/dt/codexa?color=blue&label=Downloads" alt="Downloads"></a>
   <img src="https://img.shields.io/badge/python-3.11%2B-blue" alt="Python 3.11+">
-  <img src="https://img.shields.io/badge/version-0.4.3-green" alt="Version">
-  <img src="https://img.shields.io/badge/tests-2595-brightgreen" alt="Tests">
+  <img src="https://img.shields.io/badge/version-0.5.0-green" alt="Version">
+  <img src="https://img.shields.io/badge/tests-2657-brightgreen" alt="Tests">
   <img src="https://img.shields.io/badge/coverage-79%25-brightgreen" alt="Coverage">
   <img src="https://img.shields.io/badge/mypy-strict-blue" alt="mypy strict">
   <img src="https://img.shields.io/badge/license-MIT-blue" alt="License">
@@ -84,17 +86,22 @@ structured tool protocol that any AI agent can call over HTTP or CLI.
 
 | Area | What you get |
 |------|-------------|
-| **Code Indexing** | Scan repos, extract functions/classes, generate vector embeddings (sentence-transformers + FAISS), ONNX runtime option, parallel indexing, `--watch` live re-indexing, `.codexaignore` support |
-| **Multi-Mode Search** | Semantic, keyword (BM25), regex, hybrid (RRF), and raw filesystem grep (ripgrep backend) with full `-A/-B/-C/-w/-v/-c` flags |
+| **Code Indexing** | Scan repos, extract functions/classes, generate vector embeddings (sentence-transformers + FAISS), ONNX runtime option, parallel indexing, `--watch` live re-indexing, `.codexaignore` support, `--add`/`--inspect` per-file control, model-consistency guard, Ctrl+C partial-save |
+| **Rust Search Engine** | Native `codexa-core` Rust crate via PyO3 — HNSW approximate nearest-neighbour search, BM25 keyword index, tree-sitter AST chunker (10 languages), memory-mapped vector persistence, parallel file scanner, optional ONNX embedding inference, optional Tantivy full-text search |
+| **Multi-Mode Search** | Semantic, keyword (BM25), regex, hybrid (RRF), and raw filesystem grep (ripgrep backend) with full `-A/-B/-C/-w/-v/-c/-l/-L/--exclude/--no-ignore` flags, `--hybrid`/`--sem` shorthands, `--scores`, `--snippet-length`, `--no-snippet`, JSONL streaming |
+| **RAG Pipeline** | 4-stage Retrieval-Augmented Generation — Retrieve → Deduplicate → Re-rank → Assemble with token budget, cross-encoder re-ranking, source citations |
 | **Code Context** | Rich context windows — imports, dependencies, AST-based call graphs, surrounding code |
 | **Repository Analysis** | Language breakdown (`codexa languages`), module summaries, component detection |
-| **AI Agent Protocol** | 13 built-in tools exposed via HTTP bridge, MCP server (13 tools), MCP-over-SSE (`--mcp`), or CLI for any AI agent to invoke |
+| **AI Agent Protocol** | 13 built-in tools exposed via HTTP bridge, MCP server (13 tools with pagination/cursors), MCP-over-SSE (`--mcp`), `codexa --serve` shorthand, Claude Desktop auto-config (`--claude-config`), or CLI for any AI agent to invoke |
 | **Quality & Metrics** | Complexity analysis, maintainability scoring, quality gates for CI |
 | **Multi-Repo Workspaces** | Link multiple repos under one workspace for cross-repo search & refactoring |
 | **Interactive TUI** | Terminal REPL with mode switching for interactive exploration |
 | **Streaming Responses** | Token-by-token streaming for chat and investigation commands |
 | **Plugin System** | 22 hooks for extending every layer — from indexing to tool invocation |
 | **VS Code Extension** | 4-panel sidebar (Search, Symbols, Quality, Tools), 8 commands, CodeLens, context menus, status bar |
+| **Editor Plugins** | Zed, JetBrains (IntelliJ/PyCharm), Neovim (telescope.nvim), Vim, Sublime Text, Emacs, Helix, Eclipse -- all sharing the same MCP/bridge protocol |
+| **Cross-Language Intelligence** | FFI pattern detection, polyglot dependency graphs, language-aware search boosting, universal multi-language call graph |
+| **Multi-Agent Sessions** | Concurrent AI agent sessions with shared discovery, semantic diff (rename/move/signature/body detection), RAG code generation |
 
 ---
 
@@ -140,6 +147,15 @@ cd /path/to/your-project
 codexa init
 ```
 
+CodexA auto-detects your available RAM and picks the best embedding model.
+Or choose a model profile explicitly:
+
+```bash
+codexa init --profile fast       # mxbai-embed-xsmall — low RAM (<1 GB)
+codexa init --profile balanced   # MiniLM — good balance (~2 GB)
+codexa init --profile precise    # jina-code — best quality (~4 GB)
+```
+
 This creates a `.codexa/` directory with configuration, index storage, and session data.
 
 ### 3. Index the Codebase
@@ -284,7 +300,7 @@ cd CodexA
 pip install -e ".[dev]"
 
 # Verify
-codexa --version    # → codexa, version 0.4.3
+codexa --version    # → codexa, version 0.5.0
 ```
 
 ### Step 2 — Initialize your target project
@@ -526,7 +542,7 @@ CodexA provides **39 commands** (plus subcommands) organized by capability:
 | `codexa plugin list\|scaffold\|discover` | Plugin management |
 | `codexa tui` | Interactive terminal REPL |
 | `codexa mcp` | Start MCP (Model Context Protocol) server |
-| `codexa models list\|info\|download\|switch` | Manage embedding models |
+| `codexa models list\|info\|download\|switch\|profiles\|benchmark` | Manage and benchmark embedding models |
 
 ### VS Code Extension
 
@@ -583,8 +599,13 @@ Additional tools can be registered via the plugin system using the
 │  /tools/invoke · /tools/list · /request · SSE stream │
 ├──────────────┬──────────────┬───────────────────────┤
 │ Parsing      │ Embedding    │ Search                │
-│ tree-sitter  │ sent-trans   │ FAISS                 │
+│ tree-sitter  │ sent-trans   │ FAISS / Rust HNSW     │
 ├──────────────┼──────────────┴───────────────────────┤
+│ Rust Engine  │  codexa-core (PyO3)                   │
+│ (optional)   │  HNSW · BM25 · AST chunk · mmap · RRF│
+├──────────────┼──────────────────────────────────────┤
+│ RAG Pipeline │  Retrieve → Dedup → Re-rank → Assemble│
+├──────────────┼──────────────────────────────────────┤
 │ Evolution    │  Self-improving dev loop              │
 │ engine       │  budget · task · patch · test · commit│
 ├──────────────┴──────────────────────────────────────┤
@@ -625,6 +646,9 @@ After `codexa init`, your project has `.codexa/config.json`:
 }
 ```
 
+> **Tip:** Instead of editing `model_name` manually, use `codexa init --profile fast|balanced|precise`
+> or run `codexa models profiles` to see recommended models for your hardware.
+
 ---
 
 ## Documentation
@@ -655,7 +679,7 @@ Browse the docs at **http://localhost:5173** after running `npm run docs:dev`.
 # Install dev dependencies
 pip install -e ".[dev]"
 
-# Run all 2595 tests
+# Run all 2657 tests
 pytest
 
 # Run with coverage (gate: 70% minimum)
@@ -674,10 +698,11 @@ codexa --verbose search "query"
 ## Tech Stack
 
 - **Python 3.11+** — No heavy frameworks, stdlib-first design
+- **Rust (codexa-core)** — Native search engine via PyO3 — HNSW (instant-distance), BM25, tree-sitter AST chunking, mmap persistence, parallel scanning (rayon)
 - **click** — CLI framework
 - **sentence-transformers** — Embedding generation (`all-MiniLM-L6-v2`)
-- **faiss-cpu** — Vector similarity search (O(1) file-level index, batch reconstruction)
-- **tree-sitter** — Multi-language code parsing
+- **faiss-cpu** — Vector similarity search (with Rust HNSW acceleration)
+- **tree-sitter** — Multi-language code parsing (Python + Rust)
 - **watchfiles** — Rust-backed native file watching (inotify/FSEvents/ReadDirectoryChanges)
 - **pydantic** — Configuration & data models
 - **rich** — Terminal UI and formatting

{codexa-0.4.3 → codexa-0.5.0}/README.md

@@ -5,9 +5,11 @@
 
 <p align="center">
   <a href="https://github.com/M9nx/CodexA/actions"><img src="https://github.com/M9nx/CodexA/actions/workflows/ci.yml/badge.svg" alt="CI"></a>
+  <a href="https://pypi.org/project/codexa/"><img src="https://img.shields.io/pypi/v/codexa?color=blue&label=PyPI" alt="PyPI"></a>
+  <a href="https://pepy.tech/project/codexa"><img src="https://img.shields.io/pepy/dt/codexa?color=blue&label=Downloads" alt="Downloads"></a>
   <img src="https://img.shields.io/badge/python-3.11%2B-blue" alt="Python 3.11+">
-  <img src="https://img.shields.io/badge/version-0.4.3-green" alt="Version">
-  <img src="https://img.shields.io/badge/tests-2595-brightgreen" alt="Tests">
+  <img src="https://img.shields.io/badge/version-0.5.0-green" alt="Version">
+  <img src="https://img.shields.io/badge/tests-2657-brightgreen" alt="Tests">
   <img src="https://img.shields.io/badge/coverage-79%25-brightgreen" alt="Coverage">
   <img src="https://img.shields.io/badge/mypy-strict-blue" alt="mypy strict">
   <img src="https://img.shields.io/badge/license-MIT-blue" alt="License">
@@ -24,17 +26,22 @@ structured tool protocol that any AI agent can call over HTTP or CLI.
 
 | Area | What you get |
 |------|-------------|
-| **Code Indexing** | Scan repos, extract functions/classes, generate vector embeddings (sentence-transformers + FAISS), ONNX runtime option, parallel indexing, `--watch` live re-indexing, `.codexaignore` support |
-| **Multi-Mode Search** | Semantic, keyword (BM25), regex, hybrid (RRF), and raw filesystem grep (ripgrep backend) with full `-A/-B/-C/-w/-v/-c` flags |
+| **Code Indexing** | Scan repos, extract functions/classes, generate vector embeddings (sentence-transformers + FAISS), ONNX runtime option, parallel indexing, `--watch` live re-indexing, `.codexaignore` support, `--add`/`--inspect` per-file control, model-consistency guard, Ctrl+C partial-save |
+| **Rust Search Engine** | Native `codexa-core` Rust crate via PyO3 — HNSW approximate nearest-neighbour search, BM25 keyword index, tree-sitter AST chunker (10 languages), memory-mapped vector persistence, parallel file scanner, optional ONNX embedding inference, optional Tantivy full-text search |
+| **Multi-Mode Search** | Semantic, keyword (BM25), regex, hybrid (RRF), and raw filesystem grep (ripgrep backend) with full `-A/-B/-C/-w/-v/-c/-l/-L/--exclude/--no-ignore` flags, `--hybrid`/`--sem` shorthands, `--scores`, `--snippet-length`, `--no-snippet`, JSONL streaming |
+| **RAG Pipeline** | 4-stage Retrieval-Augmented Generation — Retrieve → Deduplicate → Re-rank → Assemble with token budget, cross-encoder re-ranking, source citations |
 | **Code Context** | Rich context windows — imports, dependencies, AST-based call graphs, surrounding code |
 | **Repository Analysis** | Language breakdown (`codexa languages`), module summaries, component detection |
-| **AI Agent Protocol** | 13 built-in tools exposed via HTTP bridge, MCP server (13 tools), MCP-over-SSE (`--mcp`), or CLI for any AI agent to invoke |
+| **AI Agent Protocol** | 13 built-in tools exposed via HTTP bridge, MCP server (13 tools with pagination/cursors), MCP-over-SSE (`--mcp`), `codexa --serve` shorthand, Claude Desktop auto-config (`--claude-config`), or CLI for any AI agent to invoke |
 | **Quality & Metrics** | Complexity analysis, maintainability scoring, quality gates for CI |
 | **Multi-Repo Workspaces** | Link multiple repos under one workspace for cross-repo search & refactoring |
 | **Interactive TUI** | Terminal REPL with mode switching for interactive exploration |
 | **Streaming Responses** | Token-by-token streaming for chat and investigation commands |
 | **Plugin System** | 22 hooks for extending every layer — from indexing to tool invocation |
 | **VS Code Extension** | 4-panel sidebar (Search, Symbols, Quality, Tools), 8 commands, CodeLens, context menus, status bar |
+| **Editor Plugins** | Zed, JetBrains (IntelliJ/PyCharm), Neovim (telescope.nvim), Vim, Sublime Text, Emacs, Helix, Eclipse -- all sharing the same MCP/bridge protocol |
+| **Cross-Language Intelligence** | FFI pattern detection, polyglot dependency graphs, language-aware search boosting, universal multi-language call graph |
+| **Multi-Agent Sessions** | Concurrent AI agent sessions with shared discovery, semantic diff (rename/move/signature/body detection), RAG code generation |
 
 ---
 
@@ -80,6 +87,15 @@ cd /path/to/your-project
 codexa init
 ```
 
+CodexA auto-detects your available RAM and picks the best embedding model.
+Or choose a model profile explicitly:
+
+```bash
+codexa init --profile fast       # mxbai-embed-xsmall — low RAM (<1 GB)
+codexa init --profile balanced   # MiniLM — good balance (~2 GB)
+codexa init --profile precise    # jina-code — best quality (~4 GB)
+```
+
 This creates a `.codexa/` directory with configuration, index storage, and session data.
 
 ### 3. Index the Codebase
@@ -224,7 +240,7 @@ cd CodexA
 pip install -e ".[dev]"
 
 # Verify
-codexa --version    # → codexa, version 0.4.3
+codexa --version    # → codexa, version 0.5.0
 ```
 
 ### Step 2 — Initialize your target project
@@ -466,7 +482,7 @@ CodexA provides **39 commands** (plus subcommands) organized by capability:
 | `codexa plugin list\|scaffold\|discover` | Plugin management |
 | `codexa tui` | Interactive terminal REPL |
 | `codexa mcp` | Start MCP (Model Context Protocol) server |
-| `codexa models list\|info\|download\|switch` | Manage embedding models |
+| `codexa models list\|info\|download\|switch\|profiles\|benchmark` | Manage and benchmark embedding models |
 
 ### VS Code Extension
 
@@ -523,8 +539,13 @@ Additional tools can be registered via the plugin system using the
 │  /tools/invoke · /tools/list · /request · SSE stream │
 ├──────────────┬──────────────┬───────────────────────┤
 │ Parsing      │ Embedding    │ Search                │
-│ tree-sitter  │ sent-trans   │ FAISS                 │
+│ tree-sitter  │ sent-trans   │ FAISS / Rust HNSW     │
 ├──────────────┼──────────────┴───────────────────────┤
+│ Rust Engine  │  codexa-core (PyO3)                   │
+│ (optional)   │  HNSW · BM25 · AST chunk · mmap · RRF│
+├──────────────┼──────────────────────────────────────┤
+│ RAG Pipeline │  Retrieve → Dedup → Re-rank → Assemble│
+├──────────────┼──────────────────────────────────────┤
 │ Evolution    │  Self-improving dev loop              │
 │ engine       │  budget · task · patch · test · commit│
 ├──────────────┴──────────────────────────────────────┤
@@ -565,6 +586,9 @@ After `codexa init`, your project has `.codexa/config.json`:
 }
 ```
 
+> **Tip:** Instead of editing `model_name` manually, use `codexa init --profile fast|balanced|precise`
+> or run `codexa models profiles` to see recommended models for your hardware.
+
 ---
 
 ## Documentation
@@ -595,7 +619,7 @@ Browse the docs at **http://localhost:5173** after running `npm run docs:dev`.
 # Install dev dependencies
 pip install -e ".[dev]"
 
-# Run all 2595 tests
+# Run all 2657 tests
 pytest
 
 # Run with coverage (gate: 70% minimum)
@@ -614,10 +638,11 @@ codexa --verbose search "query"
 ## Tech Stack
 
 - **Python 3.11+** — No heavy frameworks, stdlib-first design
+- **Rust (codexa-core)** — Native search engine via PyO3 — HNSW (instant-distance), BM25, tree-sitter AST chunking, mmap persistence, parallel scanning (rayon)
 - **click** — CLI framework
 - **sentence-transformers** — Embedding generation (`all-MiniLM-L6-v2`)
-- **faiss-cpu** — Vector similarity search (O(1) file-level index, batch reconstruction)
-- **tree-sitter** — Multi-language code parsing
+- **faiss-cpu** — Vector similarity search (with Rust HNSW acceleration)
+- **tree-sitter** — Multi-language code parsing (Python + Rust)
 - **watchfiles** — Rust-backed native file watching (inotify/FSEvents/ReadDirectoryChanges)
 - **pydantic** — Configuration & data models
 - **rich** — Terminal UI and formatting

{codexa-0.4.3 → codexa-0.5.0}/codexa.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: codexa
-Version: 0.4.3
+Version: 0.5.0
 Summary: Developer intelligence CLI — semantic code search, AI-assisted understanding, and agent tooling for codebases
 License: MIT
 Project-URL: Homepage, https://codex-a.dev
@@ -65,9 +65,11 @@ Dynamic: license-file
 
 <p align="center">
   <a href="https://github.com/M9nx/CodexA/actions"><img src="https://github.com/M9nx/CodexA/actions/workflows/ci.yml/badge.svg" alt="CI"></a>
+  <a href="https://pypi.org/project/codexa/"><img src="https://img.shields.io/pypi/v/codexa?color=blue&label=PyPI" alt="PyPI"></a>
+  <a href="https://pepy.tech/project/codexa"><img src="https://img.shields.io/pepy/dt/codexa?color=blue&label=Downloads" alt="Downloads"></a>
   <img src="https://img.shields.io/badge/python-3.11%2B-blue" alt="Python 3.11+">
-  <img src="https://img.shields.io/badge/version-0.4.3-green" alt="Version">
-  <img src="https://img.shields.io/badge/tests-2595-brightgreen" alt="Tests">
+  <img src="https://img.shields.io/badge/version-0.5.0-green" alt="Version">
+  <img src="https://img.shields.io/badge/tests-2657-brightgreen" alt="Tests">
   <img src="https://img.shields.io/badge/coverage-79%25-brightgreen" alt="Coverage">
   <img src="https://img.shields.io/badge/mypy-strict-blue" alt="mypy strict">
   <img src="https://img.shields.io/badge/license-MIT-blue" alt="License">
@@ -84,17 +86,22 @@ structured tool protocol that any AI agent can call over HTTP or CLI.
 
 | Area | What you get |
 |------|-------------|
-| **Code Indexing** | Scan repos, extract functions/classes, generate vector embeddings (sentence-transformers + FAISS), ONNX runtime option, parallel indexing, `--watch` live re-indexing, `.codexaignore` support |
-| **Multi-Mode Search** | Semantic, keyword (BM25), regex, hybrid (RRF), and raw filesystem grep (ripgrep backend) with full `-A/-B/-C/-w/-v/-c` flags |
+| **Code Indexing** | Scan repos, extract functions/classes, generate vector embeddings (sentence-transformers + FAISS), ONNX runtime option, parallel indexing, `--watch` live re-indexing, `.codexaignore` support, `--add`/`--inspect` per-file control, model-consistency guard, Ctrl+C partial-save |
+| **Rust Search Engine** | Native `codexa-core` Rust crate via PyO3 — HNSW approximate nearest-neighbour search, BM25 keyword index, tree-sitter AST chunker (10 languages), memory-mapped vector persistence, parallel file scanner, optional ONNX embedding inference, optional Tantivy full-text search |
+| **Multi-Mode Search** | Semantic, keyword (BM25), regex, hybrid (RRF), and raw filesystem grep (ripgrep backend) with full `-A/-B/-C/-w/-v/-c/-l/-L/--exclude/--no-ignore` flags, `--hybrid`/`--sem` shorthands, `--scores`, `--snippet-length`, `--no-snippet`, JSONL streaming |
+| **RAG Pipeline** | 4-stage Retrieval-Augmented Generation — Retrieve → Deduplicate → Re-rank → Assemble with token budget, cross-encoder re-ranking, source citations |
 | **Code Context** | Rich context windows — imports, dependencies, AST-based call graphs, surrounding code |
 | **Repository Analysis** | Language breakdown (`codexa languages`), module summaries, component detection |
-| **AI Agent Protocol** | 13 built-in tools exposed via HTTP bridge, MCP server (13 tools), MCP-over-SSE (`--mcp`), or CLI for any AI agent to invoke |
+| **AI Agent Protocol** | 13 built-in tools exposed via HTTP bridge, MCP server (13 tools with pagination/cursors), MCP-over-SSE (`--mcp`), `codexa --serve` shorthand, Claude Desktop auto-config (`--claude-config`), or CLI for any AI agent to invoke |
 | **Quality & Metrics** | Complexity analysis, maintainability scoring, quality gates for CI |
 | **Multi-Repo Workspaces** | Link multiple repos under one workspace for cross-repo search & refactoring |
 | **Interactive TUI** | Terminal REPL with mode switching for interactive exploration |
 | **Streaming Responses** | Token-by-token streaming for chat and investigation commands |
 | **Plugin System** | 22 hooks for extending every layer — from indexing to tool invocation |
 | **VS Code Extension** | 4-panel sidebar (Search, Symbols, Quality, Tools), 8 commands, CodeLens, context menus, status bar |
+| **Editor Plugins** | Zed, JetBrains (IntelliJ/PyCharm), Neovim (telescope.nvim), Vim, Sublime Text, Emacs, Helix, Eclipse -- all sharing the same MCP/bridge protocol |
+| **Cross-Language Intelligence** | FFI pattern detection, polyglot dependency graphs, language-aware search boosting, universal multi-language call graph |
+| **Multi-Agent Sessions** | Concurrent AI agent sessions with shared discovery, semantic diff (rename/move/signature/body detection), RAG code generation |
 
 ---
 
@@ -140,6 +147,15 @@ cd /path/to/your-project
 codexa init
 ```
 
+CodexA auto-detects your available RAM and picks the best embedding model.
+Or choose a model profile explicitly:
+
+```bash
+codexa init --profile fast       # mxbai-embed-xsmall — low RAM (<1 GB)
+codexa init --profile balanced   # MiniLM — good balance (~2 GB)
+codexa init --profile precise    # jina-code — best quality (~4 GB)
+```
+
 This creates a `.codexa/` directory with configuration, index storage, and session data.
 
 ### 3. Index the Codebase
@@ -284,7 +300,7 @@ cd CodexA
 pip install -e ".[dev]"
 
 # Verify
-codexa --version    # → codexa, version 0.4.3
+codexa --version    # → codexa, version 0.5.0
 ```
 
 ### Step 2 — Initialize your target project
@@ -526,7 +542,7 @@ CodexA provides **39 commands** (plus subcommands) organized by capability:
 | `codexa plugin list\|scaffold\|discover` | Plugin management |
 | `codexa tui` | Interactive terminal REPL |
 | `codexa mcp` | Start MCP (Model Context Protocol) server |
-| `codexa models list\|info\|download\|switch` | Manage embedding models |
+| `codexa models list\|info\|download\|switch\|profiles\|benchmark` | Manage and benchmark embedding models |
 
 ### VS Code Extension
 
@@ -583,8 +599,13 @@ Additional tools can be registered via the plugin system using the
 │  /tools/invoke · /tools/list · /request · SSE stream │
 ├──────────────┬──────────────┬───────────────────────┤
 │ Parsing      │ Embedding    │ Search                │
-│ tree-sitter  │ sent-trans   │ FAISS                 │
+│ tree-sitter  │ sent-trans   │ FAISS / Rust HNSW     │
 ├──────────────┼──────────────┴───────────────────────┤
+│ Rust Engine  │  codexa-core (PyO3)                   │
+│ (optional)   │  HNSW · BM25 · AST chunk · mmap · RRF│
+├──────────────┼──────────────────────────────────────┤
+│ RAG Pipeline │  Retrieve → Dedup → Re-rank → Assemble│
+├──────────────┼──────────────────────────────────────┤
 │ Evolution    │  Self-improving dev loop              │
 │ engine       │  budget · task · patch · test · commit│
 ├──────────────┴──────────────────────────────────────┤
@@ -625,6 +646,9 @@ After `codexa init`, your project has `.codexa/config.json`:
 }
 ```
 
+> **Tip:** Instead of editing `model_name` manually, use `codexa init --profile fast|balanced|precise`
+> or run `codexa models profiles` to see recommended models for your hardware.
+
 ---
 
 ## Documentation
@@ -655,7 +679,7 @@ Browse the docs at **http://localhost:5173** after running `npm run docs:dev`.
 # Install dev dependencies
 pip install -e ".[dev]"
 
-# Run all 2595 tests
+# Run all 2657 tests
 pytest
 
 # Run with coverage (gate: 70% minimum)
@@ -674,10 +698,11 @@ codexa --verbose search "query"
 ## Tech Stack
 
 - **Python 3.11+** — No heavy frameworks, stdlib-first design
+- **Rust (codexa-core)** — Native search engine via PyO3 — HNSW (instant-distance), BM25, tree-sitter AST chunking, mmap persistence, parallel scanning (rayon)
 - **click** — CLI framework
 - **sentence-transformers** — Embedding generation (`all-MiniLM-L6-v2`)
-- **faiss-cpu** — Vector similarity search (O(1) file-level index, batch reconstruction)
-- **tree-sitter** — Multi-language code parsing
+- **faiss-cpu** — Vector similarity search (with Rust HNSW acceleration)
+- **tree-sitter** — Multi-language code parsing (Python + Rust)
 - **watchfiles** — Rust-backed native file watching (inotify/FSEvents/ReadDirectoryChanges)
 - **pydantic** — Configuration & data models
 - **rich** — Terminal UI and formatting

{codexa-0.4.3 → codexa-0.5.0}/codexa.egg-info/SOURCES.txt

@@ -8,9 +8,13 @@ codexa.egg-info/entry_points.txt
 codexa.egg-info/requires.txt
 codexa.egg-info/top_level.txt
 semantic_code_intelligence/__init__.py
+semantic_code_intelligence/rust_backend.py
 semantic_code_intelligence/version.py
 semantic_code_intelligence/analysis/__init__.py
 semantic_code_intelligence/analysis/ai_features.py
+semantic_code_intelligence/analysis/codegen.py
+semantic_code_intelligence/analysis/cross_language.py
+semantic_code_intelligence/analysis/semantic_diff.py
 semantic_code_intelligence/bridge/__init__.py
 semantic_code_intelligence/bridge/context_provider.py
 semantic_code_intelligence/bridge/protocol.py
@@ -103,12 +107,14 @@ semantic_code_intelligence/llm/mock_provider.py
 semantic_code_intelligence/llm/ollama_provider.py
 semantic_code_intelligence/llm/openai_provider.py
 semantic_code_intelligence/llm/provider.py
+semantic_code_intelligence/llm/rag.py
 semantic_code_intelligence/llm/rate_limiter.py
 semantic_code_intelligence/llm/reasoning.py
 semantic_code_intelligence/llm/safety.py
 semantic_code_intelligence/llm/streaming.py
 semantic_code_intelligence/lsp/__init__.py
 semantic_code_intelligence/mcp/__init__.py
+semantic_code_intelligence/mcp/claude_config.py
 semantic_code_intelligence/parsing/__init__.py
 semantic_code_intelligence/parsing/parser.py
 semantic_code_intelligence/plugins/__init__.py
@@ -125,6 +131,7 @@ semantic_code_intelligence/search/section_expander.py
 semantic_code_intelligence/services/__init__.py
 semantic_code_intelligence/services/indexing_service.py
 semantic_code_intelligence/services/search_service.py
+semantic_code_intelligence/sessions/__init__.py
 semantic_code_intelligence/storage/__init__.py
 semantic_code_intelligence/storage/chunk_hash_store.py
 semantic_code_intelligence/storage/hash_store.py
@@ -166,6 +173,7 @@ semantic_code_intelligence/tests/test_phase24.py
 semantic_code_intelligence/tests/test_phase25.py
 semantic_code_intelligence/tests/test_phase26.py
 semantic_code_intelligence/tests/test_phase27.py
+semantic_code_intelligence/tests/test_phase38_42.py
 semantic_code_intelligence/tests/test_phase8.py
 semantic_code_intelligence/tests/test_phase9.py
 semantic_code_intelligence/tests/test_plugins.py

{codexa-0.4.3 → codexa-0.5.0}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "codexa"
-version = "0.4.3"
+version = "0.5.0"
 description = "Developer intelligence CLI — semantic code search, AI-assisted understanding, and agent tooling for codebases"
 readme = "README.md"
 requires-python = ">=3.11"

{codexa-0.4.3 → codexa-0.5.0}/semantic_code_intelligence/analysis/ai_features.py

@@ -7,7 +7,7 @@ for feeding into LLMs or other AI systems.
 from __future__ import annotations
 
 import json
-from dataclasses import dataclass, field
+from dataclasses import asdict, dataclass, field
 from pathlib import Path
 from typing import Any
 
@@ -40,15 +40,7 @@ class LanguageStats:
 
     def to_dict(self) -> dict[str, Any]:
         """Serialize the language statistics to a plain dictionary."""
-        return {
-            "language": self.language,
-            "file_count": self.file_count,
-            "function_count": self.function_count,
-            "class_count": self.class_count,
-            "method_count": self.method_count,
-            "import_count": self.import_count,
-            "total_lines": self.total_lines,
-        }
+        return asdict(self)
 
 
 @dataclass
@@ -251,13 +243,7 @@ class CodeExplanation:
 
     def to_dict(self) -> dict[str, Any]:
         """Serialize the code explanation to a plain dictionary."""
-        return {
-            "symbol_name": self.symbol_name,
-            "symbol_kind": self.symbol_kind,
-            "file_path": self.file_path,
-            "summary": self.summary,
-            "details": self.details,
-        }
+        return asdict(self)
 
     def render(self) -> str:
         """Render the explanation as a Markdown-style string."""

codexa-0.5.0/semantic_code_intelligence/analysis/codegen.py

@@ -0,0 +1,143 @@
+"""Code generation — RAG-grounded scaffolding using codebase context.
+
+Generates code scaffolds, tests, and documentation grounded in the
+actual codebase via retrieval-augmented generation.
+"""
+
+from __future__ import annotations
+
+from dataclasses import dataclass, field
+from typing import Any
+
+from semantic_code_intelligence.utils.logging import get_logger
+
+logger = get_logger("codegen")
+
+
+@dataclass
+class CodeGenRequest:
+    """A request for code generation."""
+
+    prompt: str
+    kind: str = "scaffold"  # scaffold, test, docstring, refactor
+    target_file: str | None = None
+    target_symbol: str | None = None
+    language: str | None = None
+    max_context_chunks: int = 5
+
+    def to_dict(self) -> dict[str, Any]:
+        return {
+            "prompt": self.prompt,
+            "kind": self.kind,
+            "target_file": self.target_file,
+            "target_symbol": self.target_symbol,
+            "language": self.language,
+            "max_context_chunks": self.max_context_chunks,
+        }
+
+
+@dataclass
+class CodeGenResult:
+    """Result of a code generation request."""
+
+    generated_code: str = ""
+    context_used: list[dict[str, Any]] = field(default_factory=list)
+    model_used: str = ""
+    prompt_tokens: int = 0
+    success: bool = True
+    error: str = ""
+
+    def to_dict(self) -> dict[str, Any]:
+        return {
+            "generated_code": self.generated_code,
+            "context_used": self.context_used,
+            "model_used": self.model_used,
+            "prompt_tokens": self.prompt_tokens,
+            "success": self.success,
+            "error": self.error,
+        }
+
+
+class CodeGenerator:
+    """RAG-grounded code generator.
+
+    Retrieves relevant context from the indexed codebase, assembles a
+    prompt, and sends it to the configured LLM provider.
+    """
+
+    def __init__(self, project_root: str) -> None:
+        from pathlib import Path
+        self._project_root = Path(project_root).resolve()
+
+    def generate(self, request: CodeGenRequest) -> CodeGenResult:
+        """Generate code using RAG context + LLM."""
+        from pathlib import Path
+
+        from semantic_code_intelligence.config.settings import load_config
+        from semantic_code_intelligence.services.search_service import search_codebase
+
+        config = load_config(self._project_root)
+        result = CodeGenResult()
+
+        # Step 1: Retrieve relevant context
+        try:
+            search_results = search_codebase(
+                query=request.prompt,
+                project_root=self._project_root,
+                top_k=request.max_context_chunks,
+                mode="hybrid",
+            )
+            context_chunks = []
+            for sr in search_results:
+                context_chunks.append({
+                    "file_path": sr.file_path,
+                    "start_line": sr.start_line,
+                    "content": sr.content[:500],
+                    "score": round(sr.score, 3),
+                })
+            result.context_used = context_chunks
+        except Exception as exc:
+            logger.warning("Context retrieval failed: %s", exc)
+            context_chunks = []
+
+        # Step 2: Build the prompt
+        context_text = ""
+        for chunk in context_chunks:
+            context_text += f"\n--- {chunk['file_path']}:{chunk['start_line']} ---\n"
+            context_text += chunk["content"] + "\n"
+
+        system_prompt = (
+            "You are a code generator. Generate code based on the user's request, "
+            "grounded in the provided codebase context. Follow the existing code style "
+            "and patterns. Only output the code, no explanations."
+        )
+
+        user_prompt = f"Request: {request.prompt}\n"
+        if request.kind:
+            user_prompt += f"Type: {request.kind}\n"
+        if request.target_file:
+            user_prompt += f"Target file: {request.target_file}\n"
+        if request.language:
+            user_prompt += f"Language: {request.language}\n"
+        if context_text:
+            user_prompt += f"\nRelevant codebase context:\n{context_text}"
+
+        # Step 3: Send to LLM
+        try:
+            from semantic_code_intelligence.llm.providers import get_provider
+
+            provider = get_provider(config.llm)
+            response = provider.complete(
+                system_prompt=system_prompt,
+                user_prompt=user_prompt,
+                max_tokens=config.llm.max_tokens,
+            )
+            result.generated_code = response.content
+            result.model_used = config.llm.model
+            result.prompt_tokens = response.prompt_tokens
+        except Exception as exc:
+            result.success = False
+            result.error = str(exc)
+            logger.error("Code generation failed: %s", exc)
+
+        return result