Impact-Site-Verification: 41b53a0c-6d04-458b-a457-fe9e29acde1a

AI & Machine Learning/UnknownAI assistantmemorytoolsself-extending skills·

Moltis

Secure, self-hosted AI agent server with memory and multi-channel access

Moltis

What it does

Moltis is a persistent, self-hosted AI agent server built in Rust. It runs as a single binary on your own hardware (Mac, Linux, Windows, or Raspberry Pi) and provides a secure, sandboxed environment for an AI assistant that can remember context across sessions, execute code safely, browse the web, manage calendars, and communicate through multiple messaging channels. Unlike cloud-dependent chatbots, Moltis keeps your keys and data on your machine, with encryption-at-rest and sandboxed execution by default.

Key capabilities include:

  • Memory: Hybrid vector + full-text search for long-term context.
  • Tools: Sandboxed shell (Docker, Apple Containers, WASM), web browsing with SSRF protection, CalDAV calendar integration, cron job scheduling.
  • Self-extending skills: The agent can create its own skills, hooks, and MCP tools at runtime (Pi-inspired self-extension).
  • Multi-channel: Web UI, Telegram, WhatsApp, Discord, Slack, Matrix, Nostr, Microsoft Teams, plus GraphQL and JSON-RPC APIs.
  • Voice: Multiple cloud and local TTS/STT providers.
  • Security: Passkey/WebAuthn authentication, scoped API keys, encrypted vault, human-in-the-loop approval, no unsafe code in core paths.

Who it is for

Moltis is designed for developers, privacy-conscious users, and anyone who wants a capable AI assistant that runs on their own infrastructure. It appeals to those who:

  • Prefer self-hosting over cloud services for data control.
  • Need an AI that can interact across multiple messaging platforms from a single backend.
  • Want a secure, sandboxed environment for code execution and web browsing.
  • Value a single-binary deployment with no runtime dependencies.
  • Are comfortable with command-line installation and configuration.

Why it matters

Most AI assistants are either cloud-dependent (tying you to a provider) or require complex setups with multiple dependencies. Moltis offers a middle ground: a powerful, extensible agent that you fully own. Its security-first design—sandboxed execution, encrypted secrets, SSRF protection, and auditable Rust code—addresses real concerns about trusting AI with sensitive data. The ability to run on a Raspberry Pi or old Mac Mini makes it accessible for hobbyists, while the multi-channel support and GraphQL API make it suitable for integration into existing workflows.

Launch signal

Moltis was launched on Hacker News as a Show HN post titled "AI assistant with memory, tools, and self-extending skills." The website includes testimonials from Discord and GitHub users, and a comparison table with OpenClaw and Hermes Agent. The project is open source (MIT license) with a GitHub repository and documentation site.

Brand and naming

The name "Moltis" is short, distinctive, and easy to remember. It does not directly describe the product's function, which may reduce initial clarity but aids memorability. The brand positions itself as a secure, self-hosted alternative to cloud AI assistants, emphasizing trust and control. The tagline "A secure persistent personal agent server" clearly communicates its value proposition.

Founder

fabienpenso

Get more like this in our weekly newsletter.