Taracod's Aiden Gives Open-Source AI 89 Tools to Run Your Entire Desktop
Aiden is a fully local, open-source AI agent for Windows and Linux that controls your computer, runs 1500+ skills, and never sends data to the cloud
PRO- Local-first AI agent: Aiden runs entirely on your machine — no cloud account, no telemetry, no data leaving your hardware by default.
- Massive tool surface: 89+ built-in tools, 1500+ composable skills covering file ops, browser automation, shell execution, email, voice, and computer use.
- 6-layer memory architecture: episodic, BM25 keyword, vector semantic, procedural, goal tracking, and a permanent failure log (LESSONS.md) that grows every session.
- 14+ LLM providers: local Ollama as default with self-healing fallback chain to OpenAI, Anthropic, Groq, Cerebras, and custom OpenAI-compatible endpoints.
- MCP server mode: exposes all 89+ tools to Claude Code, Cursor, VS Code, and any MCP-compatible client as a local tool backend.
- Free and open source under AGPL-3.0 (skills are Apache-2.0); Windows gets a signed GUI installer, Linux/macOS run headless via CLI.
Most AI assistants live in a browser tab. They answer questions, forget you between sessions, and can never touch your computer. Aiden is built on a different premise: a fully local AI agent that runs as a desktop app or headless server, controls your mouse and keyboard, manages your files, browses the web, sends emails, and remembers everything across sessions , all without a cloud account or telemetry.
Aiden is a local-first AI operating system. It runs entirely on your machine , no cloud account required, no telemetry, no data leaving your hardware unless you configure a cloud provider. The project is open source under AGPL-3.0, built by a solo developer at Taracod, and is currently in active beta with weekly releases.
What it actually does
The "AI OS" framing is ambitious, but the feature list backs it up. At its core, Aiden is an autonomous agent loop that takes a plain-English goal, breaks it into steps, executes them using a large library of tools, verifies each step worked, and retries if it didn't.
- 89+ tools built in: web search, file read/write, shell execution, Playwright browser automation, screen capture and OCR, calendar, email via IMAP/SMTP, code execution sandbox, clipboard, and system info.
- 1500+ composable skills , each a self-contained directory with a
SKILL.mdprompt, tool implementations, and an optional sandbox runner , installable per-session or globally. - Computer use: Aiden can take a screenshot, understand what's on screen, and drive your mouse and keyboard to interact with any application.
- Channel adapters for Discord, Slack, Telegram, WhatsApp, Email, Webhook, and Twilio , any channel triggers the same agent loop.
- Voice I/O: speech-to-text via Groq, OpenAI, or local Whisper.cpp; text-to-speech via Edge TTS, ElevenLabs, or Windows SAPI , full offline voice loop supported.
The memory system is the real engineering story
Most agent frameworks bolt on a vector database and call it memory. Aiden uses six distinct layers, each serving a different purpose. This is worth understanding because it directly affects how the agent behaves across long sessions.
This story is for Pro members
You've reached the end of the free preview. Upgrade to AlphaSignal Pro to read the full article - and everything else behind the paywall.
Comments
