grepmax 0.7.37 → 0.7.39

package/README.md

@@ -29,44 +29,51 @@ Natural-language search that works like `grep`. Fast, local, and built for codin
 
 ## Quick Start
 
-1. **Install**
-   ```bash
-   npm install -g grepmax
-   ```
-
-2.  **Setup (Recommended)**
+```bash
+npm install -g grepmax        # 1. Install
+cd my-repo && gmax index      # 2. Index (models download automatically)
+gmax "where do we handle auth?"  # 3. Search
+```
 
-    ```bash
-    gmax setup
-    ```
+That's it. No setup required — gmax auto-detects your platform (GPU on Apple Silicon, CPU elsewhere) and downloads models on first use.
 
-    Downloads embedding models (~150MB) upfront and lets you choose between CPU (ONNX) and GPU (MLX) embedding modes. If you skip this, models download automatically on first use.
+### Optional: Interactive Setup
 
-3.  **Index**
+```bash
+gmax setup                    # Choose model tier + embedding mode interactively
+```
 
-    ```bash
-    cd my-repo
-    gmax index
-    ```
+Run this if you want to:
+- Switch between **CPU** (ONNX, works everywhere) and **GPU** (MLX, Apple Silicon only, ~3x faster)
+- Choose between **small** model (384d, 47M params, fast) and **standard** model (768d, 149M params, better quality)
 
-    Indexes into a centralized store at `~/.gmax/lancedb/`. You can index any directory — a single repo, a monorepo, or an entire workspace.
+### Quick Config (Non-Interactive)
 
-4.  **Search**
+```bash
+gmax config                          # View current settings
+gmax config --embed-mode cpu         # Switch to CPU
+gmax config --embed-mode gpu         # Switch to GPU (Apple Silicon only)
+gmax config --model-tier standard    # Switch to larger model
+```
 
-    ```bash
-    gmax "where do we handle authentication?"
-    ```
+### Verify Installation
 
-5.  **Trace** (Call Graph)
+```bash
+gmax doctor                   # Check models, index, servers
+```
 
-    ```bash
-    gmax trace "function_name"
-    ```
-    See who calls a function (upstream dependencies) and what it calls (downstream dependencies).
+### Core Commands
 
-    ```bash
-    gmax symbols                  # List all indexed symbols
-    ```
+```bash
+gmax "where do we handle auth?"      # Semantic search
+gmax "VectorDB" --symbol --agent     # Search + call graph (compact output)
+gmax trace handleAuth -d 2           # Call graph (2-hop)
+gmax skeleton src/lib/search/        # File structure (directory)
+gmax project                         # Project overview
+gmax related src/lib/syncer.ts       # Dependencies + dependents
+gmax recent                          # Recently modified files
+gmax symbols auth                    # List indexed symbols
+```
 
 In our public benchmarks, `grepmax` can save about 20% of your LLM tokens and deliver a 30% speedup.
 
@@ -82,6 +89,8 @@ In our public benchmarks, `grepmax` can save about 20% of your LLM tokens and de
 2. Open Claude Code — the plugin auto-starts the MLX GPU server and a background file watcher.
 3. Claude uses `gmax` for semantic searches automatically via MCP tools.
 
+Plugin files (skill instructions, hooks) auto-update when you run `npm update -g grepmax` — no need to re-run `install-claude-code`.
+
 ### Opencode
 1. Run `gmax install-opencode`
 2. OC uses `gmax` for semantic searches via MCP.

package/dist/commands/setup.js

@@ -118,30 +118,12 @@ exports.setup = new commander_1.Command("setup")
         return;
     }
     const selectedTier = config_1.MODEL_TIERS[modelTier];
-    // Step 5: Embed mode selection
-    const embedMode = yield p.select({
-        message: "Embedding mode",
-        options: [
-            {
-                value: "cpu",
-                label: "CPU only",
-                hint: "ONNX — works everywhere",
-            },
-            {
-                value: "gpu",
-                label: "GPU (MLX)",
-                hint: "Apple Silicon only, faster indexing + search",
-            },
-        ],
-        initialValue: (_c = existingConfig === null || existingConfig === void 0 ? void 0 : existingConfig.embedMode) !== null && _c !== void 0 ? _c : (process.arch === "arm64" && process.platform === "darwin"
-            ? "gpu"
-            : "cpu"),
-    });
-    if (p.isCancel(embedMode)) {
-        p.cancel("Setup cancelled");
-        yield (0, exit_1.gracefulExit)();
-        return;
-    }
+    // Step 5: Embed mode — auto-detect, no prompt needed
+    const isAppleSilicon = process.arch === "arm64" && process.platform === "darwin";
+    const embedMode = (_c = existingConfig === null || existingConfig === void 0 ? void 0 : existingConfig.embedMode) !== null && _c !== void 0 ? _c : (isAppleSilicon ? "gpu" : "cpu");
+    p.log.info(isAppleSilicon
+        ? "Apple Silicon detected — using GPU acceleration (MLX)"
+        : "Using CPU embeddings (ONNX)");
     const mlxModel = embedMode === "gpu" ? selectedTier.mlxModel : undefined;
     // Step 6: Write configs
     (0, index_config_1.writeSetupConfig)(paths.configPath, {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "grepmax",
-  "version": "0.7.37",
+  "version": "0.7.39",
   "author": "Robert Owens <robowens@me.com>",
   "homepage": "https://github.com/reowens/grepmax",
   "bugs": {

package/plugins/grepmax/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "grepmax",
-  "version": "0.7.37",
+  "version": "0.7.39",
   "description": "Semantic code search for Claude Code. Automatically indexes your project and provides intelligent search capabilities.",
   "author": {
     "name": "Robert Owens",