nex-code 0.3.5 → 0.3.9

package/README.md

@@ -18,7 +18,7 @@
   <a href="https://opensource.org/licenses/MIT"><img src="https://img.shields.io/badge/License-MIT-blue.svg" alt="License: MIT"></a>
   <img src="https://img.shields.io/badge/node-%3E%3D18-brightgreen.svg" alt="Node >= 18">
   <img src="https://img.shields.io/badge/dependencies-2-green.svg" alt="Dependencies: 2">
-  <img src="https://img.shields.io/badge/tests-1783-blue.svg" alt="Tests: 1783">
+  <img src="https://img.shields.io/badge/tests-1752-blue.svg" alt="Tests: 1752">
 </p>
 
 ---
@@ -47,6 +47,21 @@ That's it. You'll see the banner, your project context, and the `>` prompt. Star
 
 ---
 
+## Automatic Updates
+
+Nex Code automatically checks for new versions when you start it. If a newer version is available, you'll see a notification with instructions on how to update:
+
+```
+💡 New version available! Run npm update -g nex-code to upgrade from x.x.x to x.x.x
+```
+
+To update to the latest version:
+```bash
+npm update -g nex-code
+```
+
+---
+
 ## Why nex-code?
 
 | | nex-code | aider | Cursor |
@@ -55,7 +70,7 @@ That's it. You'll see the banner, your project context, and the `>` prompt. Star
 | **Multi-provider** | 5 providers, swap at runtime | Multi-provider | Multi-provider |
 | **Free local models** | Ollama (no API key) | Ollama | — |
 | **Runtime dependencies** | 2 (axios, dotenv) | Heavy | Electron |
-| **Test coverage** | 1783 tests, 90% coverage | — | — |
+| **Test coverage** | 1752 tests, 85% coverage | — | — |
 | **Tool tiers** | Auto-adapts tools per model | Fixed | Fixed |
 | **No lock-in** | `/model openai:gpt-4o` ↔ `/model local:llama3` | Config change | Config change |
 
@@ -85,6 +100,7 @@ npx nex-code
 git clone https://github.com/hybridpicker/nex-code.git
 cd nex-code
 npm install
+npm run build         # Build the high-performance bundle
 cp .env.example .env
 npm link
 npm run install-hooks
@@ -96,7 +112,7 @@ Create a `.env` file in your project directory (or set environment variables):
 
 ```bash
 # Pick any — only one is required
-OLLAMA_API_KEY=your-key       # Ollama Cloud (Qwen3 Coder, DeepSeek R1, Devstral, Kimi K2.5, Llama 4, MiniMax M2.5, GLM 4.7)
+OLLAMA_API_KEY=your-key       # Ollama Cloud (Qwen3 Coder, Qwen3.5, DeepSeek R1, Devstral, Kimi K2.5, Llama 4, MiniMax M2.5, GLM 4.7)
 OPENAI_API_KEY=your-key       # OpenAI (GPT-4o, GPT-4.1, o1, o3, o4-mini)
 ANTHROPIC_API_KEY=your-key    # Anthropic (Claude Sonnet 4.6, Opus 4.6, Haiku 4.5)
 GEMINI_API_KEY=your-key       # Google Gemini (3.1 Pro Preview, 2.5 Pro/Flash, 2.0 Flash)
@@ -193,10 +209,10 @@ Switch providers and models at runtime:
 
 | Provider | Models | Env Variable |
 |----------|--------|-------------|
-| **ollama** | Qwen3 Coder, DeepSeek R1, Devstral, Kimi K2.5, MiniMax M2.5, GLM 4.7, Llama 4 | `OLLAMA_API_KEY` |
+| **ollama** | Qwen3 Coder, Qwen3.5 (397B, 122B-A10B, 35B-A3B, 27B, 9B, 4B, 2B, 0.8B), DeepSeek R1, Devstral, Kimi K2.5, MiniMax M2.5, GLM 4.7, Llama 4 | `OLLAMA_API_KEY` |
 | **openai** | GPT-4o, GPT-4.1, o1, o3, o4-mini | `OPENAI_API_KEY` |
 | **anthropic** | Claude Sonnet 4.6, Opus 4.6, Haiku 4.5, Sonnet 4.5, Sonnet 4 | `ANTHROPIC_API_KEY` |
-| **gemini** | Gemini 3.1 Pro Preview, 2.5 Pro/Flash/Lite, 2.0 Flash/Lite | `GEMINI_API_KEY` |
+| **gemini** | Gemini 3.1 Pro Preview, 2.5 Pro/Flash, 1.5 Pro/Flash | `GEMINI_API_KEY` |
 | **local** | Any model on your local Ollama server | (none) |
 
 Fallback chains let you auto-switch when a provider fails:
@@ -299,6 +315,11 @@ When the model runs tools but produces no visible text, an automatic nudge force
 ### Response Quality
 The system prompt enforces substantive responses: the model always presents findings as formatted text after using tools (users only see 1-line tool summaries). Responses use markdown with headers, bullet lists, and code blocks. The model states its approach before non-trivial tasks and summarizes results after completing work.
 
+### Performance
+- **Asynchronous I/O**: The entire CLI is built on non-blocking I/O. File reads, writes, and git operations never block the main thread, keeping the UI responsive even during heavy tasks.
+- **Fast Startup**: Pre-bundled with `esbuild` to minimize module loading overhead, achieving sub-100ms startup times.
+- **In-Memory Indexing**: A background indexing engine (using `ripgrep` or a fast fallback) keeps project file paths in RAM for instant file discovery, path auto-fixing, and glob searches.
+
 ### Streaming Output
 Tokens appear live as the model generates them. Braille spinner during connection, then real-time line-by-line rendering via `StreamRenderer` with markdown formatting and syntax highlighting (JS, TS, Python, Go, Rust, CSS, HTML, and more).
 
@@ -310,7 +331,8 @@ Pressing Ctrl+C during a running request immediately cancels the active HTTP str
 - An `AbortController` signal flows from the SIGINT handler through the agent loop to the provider's HTTP request
 - All providers (Ollama, OpenAI, Anthropic, Gemini, local) destroy the response stream on abort
 - No EPIPE errors after cancellation (stdout writes are EPIPE-guarded)
-- 3x rapid Ctrl+C force-exits the process
+- 2x Ctrl+C during processing force-exits the process
+- At the idle prompt: first Ctrl+C shows `(Press Ctrl+C again to exit)`, second Ctrl+C exits (hint resets after 2 s)
 
 ### Diff Preview
 Every file change is shown in a diff-style format before being applied:
@@ -475,6 +497,12 @@ Four features that make Nex Code significantly more reliable with open-source mo
 - **Edit auto-fix**: Close match (≤5% Levenshtein distance) in `edit_file`/`patch_file` is auto-applied instead of erroring. Stacks with fuzzy whitespace matching.
 - **Bash error hints**: Enriches error output with actionable hints — "command not found" → install suggestion, `MODULE_NOT_FOUND` → `npm install <pkg>`, port in use, syntax errors, TypeScript errors, test failures, and more.
 
+**Stale Stream Recovery** — Progressive retry strategy when streams stall (common with large Ollama models after many agent steps):
+- 1st retry: 3s backoff delay, resend same context (handles transient stalls)
+- 2nd retry: force-compress conversation (~80k tokens freed), 5s delay, retry with smaller context
+- Last resort: if retries exhausted, one final force-compress + reset for fresh attempts
+- Broader context-too-long detection catches Ollama-specific error formats (`num_ctx`, `prompt`, `size`, `exceeds`)
+
 **Tool Tiers** — Dynamically reduces the tool set based on model capability:
 - **essential** (5 tools): bash, read_file, write_file, edit_file, list_directory
 - **standard** (13 tools): + search_files, glob, grep, ask_user, git_status, git_diff, git_log, task_list
@@ -635,6 +663,7 @@ cli/
 ├── tool-validator.js    # Tool argument validation + auto-correction
 ├── tool-tiers.js        # Dynamic tool set selection per model + model tier lookup
 ├── ui.js                # ANSI colors, banner + re-exports from format.js/spinner.js
+├── index-engine.js      # In-memory file index (ripgrep/fallback)
 ├── auto-fix.js          # Path resolution, edit matching, bash error hints
 ├── tool-retry.js        # Malformed argument retry with schema hints
 └── ollama.js            # Backward-compatible wrapper
@@ -682,6 +711,29 @@ Project-local configuration and state (gitignored):
 
 ---
 
+## Performance
+
+Nex Code v0.3.7+ includes comprehensive performance optimizations:
+
+| Optimization | Improvement | Impact |
+|--------------|-------------|--------|
+| **System Prompt Caching** | 4.3× faster | 77µs → 18µs |
+| **Token Estimation Caching** | 3.5× faster | Cached after first call |
+| **Context File Caching** | 10-20× faster | 50-200ms → 5-10ms |
+| **Debounced Auto-Save** | 0ms in hot path | Saves after 5s inactivity |
+| **Tool Filter Caching** | 1.7× faster | Cached per model |
+| **Schema Cache** | 3.4× faster | 2.51µs → 0.73µs |
+
+**Average speedup:** 2.7× (micro-benchmarks)  
+**Real-world improvement:** ~10× faster per turn
+
+Run benchmarks yourself:
+```bash
+node benchmark.js
+```
+
+---
+
 ## Testing
 
 ```bash
@@ -689,7 +741,7 @@ npm test              # Run all tests with coverage
 npm run test:watch    # Watch mode
 ```
 
-43 test suites, 1783 tests, 90% statement / 83% branch coverage.
+44 test suites, 1752 tests, 85% statement / 79% branch coverage.
 
 CI runs on GitHub Actions (Node 18/20/22).