Moltis
Self-hosted Rust-native AI assistant distributed as a single binary
Overview
Moltis is a self-hosted AI assistant built in Rust, distributed as a single binary with no runtime dependencies. Run it locally on your own hardware for complete privacy and control. Supports local LLMs with automatic setup alongside cloud providers like OpenAI and GitHub Copilot. Features sandboxed code execution via Docker and Apple Containers, a web UI, Telegram bot, JSON-RPC API, and mobile PWA. Includes 15+ text-to-speech and speech-to-text providers, a hybrid memory system with vector and full-text search, and MCP server support via stdio or HTTP/SSE. Security-first with passkeys/WebAuthn, SSRF protection, HTTPS by default, and secret redaction.
Architecture: Apple Silicon, Intel
Key Features
- Single binary deployment with no runtime dependencies
- Local LLM support with automatic setup and offline operation
- Sandboxed code execution via Docker and Apple Containers
- Multi-channel access: Web UI, Telegram bot, JSON-RPC API, and mobile PWA
- 15+ text-to-speech and speech-to-text providers
- Hybrid memory system with vector and full-text search
- MCP server support via stdio or HTTP/SSE
- Provider fallback chains with per-provider metrics
- Passkeys/WebAuthn authentication with HTTPS by default
