auvrynt 1.0.1

package/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 EN-rain
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

package/README.md

@@ -0,0 +1,239 @@
+
+
+<h1 align="center">Auvrynt — Web-to-local Arsenal</h1>
+
+<p align="center">
+  A self-hosted MCP server that gives AI coding assistants secure, real-time access to your local machine.
+</p>
+
+<p align="center">
+  <a href="https://www.npmjs.com/package/auvrynt">
+    <img alt="npm" src="https://img.shields.io/npm/v/auvrynt?style=flat-square&color=cb3837" />
+  </a>
+  <a href="https://github.com/EN-rain/Auvrynt-Web-to-local-Arsenal/actions/workflows/ci.yml">
+    <img alt="CI" src="https://img.shields.io/github/actions/workflow/status/EN-rain/Auvrynt-Web-to-local-Arsenal/ci.yml?style=flat-square&branch=main" />
+  </a>
+  <a href="https://github.com/EN-rain/Auvrynt-Web-to-local-Arsenal/blob/main/LICENSE">
+    <img alt="License" src="https://img.shields.io/npm/l/auvrynt?style=flat-square" />
+  </a>
+  <a href="https://nodejs.org">
+    <img alt="Node" src="https://img.shields.io/badge/node-%3E%3D20.12-brightgreen?style=flat-square" />
+  </a>
+</p>
+
+---
+
+## What is Auvrynt?
+
+Auvrynt is a **Model Context Protocol (MCP) server** you run on your own machine. It creates a secure, tunneled connection between AI clients like ChatGPT or Claude and your local development environment — your actual files, terminals, build tools, and project folders.
+
+No cloud upload. No third-party storage. Your code stays on your machine.
+
+Connect once, then ask your AI to:
+
+- Open any approved project folder on your machine
+- Read, write, search, and edit real source files
+- Run shell commands, tests, builds, and package scripts
+- Work across web, .NET, Godot, and Blender projects
+- Capture screenshots of running apps
+- Use isolated Git worktrees for parallel coding sessions
+
+---
+
+## Quick Start
+
+> **Requires:** Node.js `>=20.12 <27` — Node 22 LTS recommended
+
+**Install globally:**
+
+```bash
+npm install -g auvrynt
+```
+
+**Set up and start:**
+
+```bash
+auvrynt init
+auvrynt start
+```
+
+**Or run directly without installing:**
+
+```bash
+npx auvrynt init
+npx auvrynt start
+```
+
+During `auvrynt init`, you will be asked for:
+
+| Prompt | Description |
+|---|---|
+| Project roots | Local folder paths your AI is allowed to access |
+| Local port | Default is `49321` |
+| Public base URL | Your tunnel URL (Cloudflare, ngrok, Pinggy, Tailscale, etc.) |
+
+> **Important:** Provide the base URL — do **not** include `/mcp`. Example: `https://your-tunnel.trycloudflare.com`
+
+---
+
+## Start the Server
+
+```bash
+# Clean animated mode (shows public URL + password, with live log indicator)
+auvrynt start
+
+# Verbose mode (shows full request logs in terminal)
+auvrynt serve
+```
+
+After startup, the terminal shows:
+
+```
+Auvrynt is running!
+
+  Public URL:     https://your-tunnel.trycloudflare.com/mcp
+  Owner Password: <your-generated-password>
+
+  # CTRL + C to stop
+
+  ⠋ Logs active... (0 requests handled | Last: Started successfully)
+```
+
+---
+
+## Connect an MCP Client
+
+Paste your `/mcp` URL into your AI client (ChatGPT, Claude, etc.):
+
+```
+https://your-tunnel.trycloudflare.com/mcp
+```
+
+When the client connects, you will be prompted to approve the session using the **Owner password** shown at startup. The password is also stored in:
+
+```
+~/.auvrynt/auth.json
+```
+
+Keep it private.
+
+---
+
+## Commands
+
+| Command | Description |
+|---|---|
+| `auvrynt init` | Run first-time setup or update your config |
+| `auvrynt start` | Start the server with clean animated output |
+| `auvrynt serve` | Start the server with verbose log output |
+| `auvrynt doctor` | Show your config, Node version, and dependency health |
+| `auvrynt config get` | Print your saved configuration |
+| `auvrynt config set publicBaseUrl <url>` | Update your public tunnel URL |
+
+**Override the tunnel URL for a one-off run:**
+
+```bash
+AUVRYNT_PUBLIC_BASE_URL=https://new-tunnel.example.com auvrynt start
+```
+
+---
+
+## What Your AI Can Do
+
+Once connected and a workspace is opened, your AI assistant has access to **139 tools** across:
+
+| Category | Capabilities |
+|---|---|
+| **Files** | Read, write, edit, search, glob, list directory |
+| **Shell** | Run commands, tests, builds, git, package scripts |
+| **Processes** | Start/stop persistent processes, tail logs |
+| **Web** | Launch dev servers, screenshot pages, inspect DOM |
+| **Images** | Inspect, compare, sprite-sheet splitting |
+| **.NET** | Inspect projects, restore, build, test, format |
+| **Godot** | Run projects, inspect scenes, C# build, GDScript diagnostics |
+| **Blender** | Scene audit, checkpoints, renders, export GLB |
+| **Git** | Worktree management, status, diff, branch ops |
+
+---
+
+## Project Routing
+
+Auvrynt automatically selects the right workflow guide based on your project type:
+
+| Project signals | Guide loaded |
+|---|---|
+| `package.json` + `vite.config.ts` / web framework | Web Agent Guide |
+| `App.sln` / `.csproj` without `project.godot` | Software Agent Guide |
+| `project.godot` (any language) | Godot Agent Guide |
+| `.blend` files | Blender Agent Guide |
+
+See [`PROJECT_ROUTER.md`](guides/PROJECT_ROUTER.md) for full routing rules.
+
+---
+
+## Platform Support
+
+| Platform | Status |
+|---|---|
+| Linux | ✅ Supported |
+| macOS | ✅ Supported |
+| Windows (Git Bash / WSL / MSYS2) | ✅ Supported |
+| Windows PowerShell / cmd.exe only | ❌ Not supported — use Git Bash or WSL |
+
+**Check your environment:**
+
+```bash
+auvrynt doctor
+```
+
+---
+
+## Configuration
+
+Config files are stored in `~/.auvrynt/`:
+
+```
+~/.auvrynt/config.json   ← ports, roots, public URL
+~/.auvrynt/auth.json     ← owner password (keep private)
+```
+
+Key environment variables:
+
+| Variable | Description |
+|---|---|
+| `AUVRYNT_ALLOWED_ROOTS` | Comma-separated project root paths |
+| `AUVRYNT_PUBLIC_BASE_URL` | Your public tunnel origin (no `/mcp`) |
+| `AUVRYNT_OAUTH_OWNER_TOKEN` | Override the owner password directly |
+| `AUVRYNT_LOG_FORMAT` | `json` (default) or `pretty` |
+
+See [Configuration Reference](docs/configuration.md) for all options.
+
+---
+
+## Documentation
+
+| Document | Description |
+|---|---|
+| [Setup Guide](docs/setup.md) | Step-by-step tunnel and client setup |
+| [Configuration Reference](docs/configuration.md) | All env vars and config options |
+| [ChatGPT Coding Workflow](docs/chatgpt-coding-workflow.md) | How to use Auvrynt day-to-day |
+| [Security Model](docs/security.md) | Path isolation, auth, and threat model |
+| [Troubleshooting Gotchas](docs/gotchas.md) | Common issues and fixes |
+
+---
+
+## Local Development
+
+```bash
+npm install --include=dev
+npm run typecheck
+npm test
+npm run build
+npm run dev
+```
+
+---
+
+## License
+
+MIT — © 2026 [EN-rain](https://github.com/EN-rain)

package/dist/blender-client.js

@@ -0,0 +1,109 @@
+import { createConnection } from "node:net";
+export class BlenderClient {
+    static MAX_RESPONSE_BYTES = 8 * 1024 * 1024;
+    host;
+    port;
+    timeoutMs;
+    longTaskTimeoutMs;
+    activeMutationPromise = Promise.resolve();
+    constructor(config = {}) {
+        this.host = config.host || "127.0.0.1";
+        this.port = config.port || 49323;
+        this.timeoutMs = config.timeoutMs || 180000;
+        this.longTaskTimeoutMs = config.longTaskTimeoutMs || 7200000;
+        // Security check: enforce loopback host
+        if (this.host !== "127.0.0.1" && this.host !== "localhost") {
+            throw new Error(`Forbidden host: ${this.host}. Only loopback connection is allowed for Blender MCP.`);
+        }
+    }
+    /**
+     * Sends Python code to the Blender bridge over TCP socket.
+     * Ensures single execution at a time to prevent concurrency issues.
+     */
+    async sendExecute(code, strictJson = true, isLongTask = false) {
+        const action = () => this.sendExecuteInternal(code, strictJson, isLongTask);
+        this.activeMutationPromise = this.activeMutationPromise.then(action, action);
+        return this.activeMutationPromise;
+    }
+    async sendExecuteInternal(code, strictJson, isLongTask) {
+        const timeout = isLongTask ? this.longTaskTimeoutMs : this.timeoutMs;
+        const request = {
+            type: "execute",
+            code,
+            strict_json: strictJson,
+        };
+        const payload = Buffer.concat([Buffer.from(JSON.stringify(request), "utf8"), Buffer.from([0])]);
+        // Limit payload size to 1 MB
+        if (payload.length > 1024 * 1024) {
+            throw new Error("Payload size limit exceeded (max 1 MB)");
+        }
+        return new Promise((resolve, reject) => {
+            let client = null;
+            let buffer = Buffer.alloc(0);
+            let timeoutId = null;
+            const cleanup = () => {
+                if (timeoutId) {
+                    clearTimeout(timeoutId);
+                    timeoutId = null;
+                }
+                if (client) {
+                    client.destroy();
+                    client = null;
+                }
+            };
+            try {
+                client = createConnection({ host: this.host, port: this.port }, () => {
+                    client?.write(payload);
+                });
+                timeoutId = setTimeout(() => {
+                    cleanup();
+                    reject(new Error(`Blender connection timed out after ${timeout}ms`));
+                }, timeout);
+                client.on("data", (chunk) => {
+                    buffer = Buffer.concat([buffer, Buffer.isBuffer(chunk) ? chunk : Buffer.from(chunk)]);
+                    if (buffer.length > BlenderClient.MAX_RESPONSE_BYTES) {
+                        cleanup();
+                        reject(new Error("Blender response exceeded the 8 MB limit"));
+                        return;
+                    }
+                    // Check for null terminator
+                    const nullIndex = buffer.indexOf(0);
+                    if (nullIndex !== -1) {
+                        const raw = buffer.subarray(0, nullIndex).toString("utf8").trim();
+                        cleanup();
+                        try {
+                            const parsed = JSON.parse(raw);
+                            if (parsed.status === "ok") {
+                                resolve(parsed.result || {});
+                            }
+                            else {
+                                reject(new Error(parsed.message || "Unknown Blender execution error"));
+                            }
+                        }
+                        catch (e) {
+                            reject(new Error(`Failed to parse Blender JSON response: ${e.message}`));
+                        }
+                    }
+                });
+                client.on("error", (err) => {
+                    cleanup();
+                    reject(new Error(`Blender connection error: ${err.message}`));
+                });
+            }
+            catch (err) {
+                cleanup();
+                reject(err);
+            }
+        });
+    }
+}
+// Global active client registry matching workspace Id
+const activeBlenderClients = new Map();
+export function getBlenderClient(workspaceId, config) {
+    let client = activeBlenderClients.get(workspaceId);
+    if (!client) {
+        client = new BlenderClient(config);
+        activeBlenderClients.set(workspaceId, client);
+    }
+    return client;
+}