Repolode
Submit
dosco

dosco/aithy

A private, local-first personal AI agent runtime that runs sandboxed tools, keeps memory and knowledge on your machine, and works with local, cloud, or Grok models.

View on GitHubWebsite
Stars
108
Forks
8
Open issues
1
Last commit
2 months ago

What is it?

Aithy is a personal AI agent runtime that runs on your own machine. You start it, open the browser control surface at 127.0.0.1:3000, and use an agent workspace with sessions, memory, skills, background attentions, and artifacts. The model source is a runtime choice: managed local llama.cpp models, cloud or API providers (OpenAI, OpenRouter, Groq, Azure OpenAI and others through Ax profiles), or an eligible Grok subscription signed in with OAuth.

Command execution happens in a Microsandbox Linux sandbox, with host files exposed only through explicit attachments or mounts. Optional host shell access goes through permission prompts. Sessions, memory, knowledge bundles, usage records and settings are stored in SQLite under the local config directory. Knowledge and memory retrieval combine SQLite FTS with sqlite-vec semantic search and local reranking.

Aithy Mesh lets a laptop Aithy pair with an Aithy on a GPU machine on the same LAN. Paired peers share validated inference and search services over pinned TLS, and provider API keys are not sent to peers. The project is written in TypeScript and runs on Bun, with packaged builds for Apple Silicon macOS and Linux x64 and ARM64.

Key features

  • 01Browser control surface for chat, sessions, memory, skills and runtime services.
  • 02Runs agent commands in a Microsandbox Linux sandbox with explicit file mounts.
  • 03Supports local llama.cpp models, cloud/API providers and Grok subscription sign-in.
  • 04Stores memory, knowledge and usage data locally in SQLite with hybrid search.
  • 05Shares inference and search services between machines on a LAN through Mesh pairing.
  • 06Connects to MCP servers and exposes a read-only, bearer-authenticated MCP endpoint.

Use cases

  • Running a private personal assistant that keeps conversation memory on your own computer.
  • Using a laptop front end while sending heavy local model work to a GPU machine on the same network.
  • Running document, OCR and media-processing tasks in an isolated sandbox.
  • Curating reusable knowledge bundles and reviewing agent-proposed changes before they are saved.

Topics

#ai-agent#personal-ai#local-ai#llm#sandbox#mcp#typescript#self-hosted#ai

FAQ

What is aithy?
A private, local-first personal AI agent runtime that runs sandboxed tools, keeps memory and knowledge on your machine, and works with local, cloud, or Grok models. Aithy is a personal AI agent runtime that runs on your own machine. You start it, open the browser control surface at 127.0.0.1:3000, and use an agent workspace with sessions, memory, skills, background attentions, and artifacts. The model source is a runtime choice: managed local llama.cpp models, cloud or API providers (OpenAI, OpenRouter, Groq, Azure OpenAI and others through Ax profiles), or an eligible Grok subscription signed in with OAuth.
Is aithy free to use? What license does it have?
aithy is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
Is aithy still maintained?
Updated within 6 months. The last commit to aithy was on Aug 31, 2026.
What can I use aithy for?
Running a private personal assistant that keeps conversation memory on your own computer. Using a laptop front end while sending heavy local model work to a GPU machine on the same network. Running document, OCR and media-processing tasks in an isolated sandbox. Curating reusable knowledge bundles and reviewing agent-proposed changes before they are saved.
Where can I get aithy?
The source code and installation instructions are on GitHub: https://github.com/dosco/aithy

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects