chatbox
Local-first multi-model AI chat client with built-in IDE, PC agent, and production security hardening.
Ficha actualizada 22 sept 2026, 20:51 · upstream 22 sept 2026, 20:47 · release v1.0.0
Qué es chatbox
Descripción editorial construida a partir del repositorio, su documentación y sus señales de mantenimiento.
Chatbox is a self-contained local-first workspace that runs entirely on your own computer. Unlike public chat apps where prompts and keys live on third-party servers, Chatbox keeps everything inside one folder: the server binds to localhost, API keys never leave your machine, and all chat history, cache, and sessions are stored locally.
The application also ships a real code editor powered by Monaco (the same engine as VS Code), a live terminal that streams shell commands over SSE, and an autonomous agent capable of writing, running, and verifying code on your machine. You can switch between providers like OpenAI, Claude, Google Gemini, or Ollama without installing anything — just paste the API key.
It is designed for privacy-conscious developers who want to write, test, and run AI-powered tasks entirely offline, with full control over what runs on their system.
- Run local LLMs (Ollama, LM Studio) without needing an internet connection or API keys
- Write complete code in a browser-based IDE while running it locally
- Automate repetitive tasks with a secure agent that can execute commands on your machine
- Build and test AI-powered applications entirely offline using local models
- Share private conversations between team members without exposing data to the cloud
- Run multi-model chat (OpenAI, Anthropic, Google, Cohere) from a single interface
- Runs entirely locally with no API keys leaving your machine — all prompts and history stay on your PC
- Includes Monaco editor (VS Code engine) that works offline without any dependencies
- Built-in autonomous agent that writes, runs code, verifies results, and updates a live to-do checklist
- 5-layer security architecture including scrypt PIN verification, HMAC-signed sessions, and symlink-safe path confinement
- Supports multiple providers (OpenAI, Claude, Google, Ollama) with automatic provider detection from the API URL
- Requires a local machine to run — no cloud-based alternative available
- Local models like Ollama require you to install and maintain them on your own hardware
- The autonomous agent requires manual approval for each action, which can be slow for large tasks
- No built-in chat history persistence across sessions without local storage setup
- Terminal output is not fully searchable in the IDE by default
Perfil de infraestructura
Estimación inicial derivada del repositorio y de los servicios detectados.
Despliegue detectado
Qué señales reales de instalación hemos encontrado.
Run with `start-app.bat` for a secure localhost-only mode. For LAN access, use `start-server.bat` and open from your phone at http://<PC-IP>:3000. The app bundles its own Node.js runtime and Python interpreter — no system-wide installation needed. Update via `git pull`, then re-run the launcher. Always close the console window before stopping to prevent memory leaks.
Stack y dependencias
Componentes inferidos a partir de manifests, configuración y documentación.
Puede sustituir a
Relaciones detectadas por el análisis del repositorio y su categoría.
Experiencia real de la comunidad
Ayuda a contrastar la estimación automática con instalaciones reales.
Todavía no hay datos comunitarios para este proyecto.