askimo-ai/askimo

AI desktop app for chat, RAG, Skills, MCP tools, and agents. Support multiple LLMs (Anthropic, OpenAI, VertexAI, vLLM, Nvidia NIM, Gemini, OpenRouter)

Askimo – Desktop AI Toolkit

One app, every model. Your files stay local.

Askimo is a native desktop application (macOS, Windows, Linux) that lets you chat with a wide range of LLM providers—OpenAI, Claude, Gemini, Grok, Ollama, LM Studio, Docker AI, OpenRouter, NVIDIA NIM, Together AI, vLLM, or any OpenAI‑compatible endpoint—without opening a browser. It bundles a set of productivity‑focused AI features while keeping all data (conversations, RAG indexes, telemetry) on your machine.

Core Capabilities

Category	What it does
Multi‑provider chat	Switch instantly between cloud and local models; attach images for multimodal models.
Local RAG	Index folders, files, or URLs; hybrid BM25 + vector retrieval with an AI classifier that skips retrieval when unnecessary.
Web search	DuckDuckGo (no key), Brave Search API, Tavily, Serply, or a self‑hosted SearxNG instance.
Agentic workflows (Plans)	Build multi‑step pipelines via a form UI; steps chain automatically, live progress shown, export to PDF/Word; define plans in YAML or generate them in plain English.
Script runner	Execute Python, Bash, or JavaScript from chat; Python runs in an auto‑managed virtualenv with on‑the‑fly dependency install.
MCP integration	Connect to any MCP‑compatible server via stdio or HTTP.
AI coding agents	Run autonomous agents such as Claude Code, OpenAI Codex, or Google Antigravity; see live tool calls and reasoning.
Reusable skills	Define a skill once and make it available to any supported agent’s native skill folder.
Voice	Offline speech‑to‑text (whisper.cpp) and text‑to‑speech (Piper) or cloud APIs; optional “conversation mode” for hands‑free use.
Privacy‑by‑design	All conversation history, RAG indexes, and telemetry are stored locally in SQLite; no data is sent unless you explicitly call a provider.
Internationalisation	UI available in 10+ languages.

Getting Started

Download the installer for macOS, Windows, or Linux from the release page.
Add a provider – paste an API key (OpenAI, Claude, Gemini, etc.) or point to a running Ollama server.
Start chatting – you can immediately use the chat window, run a script, or launch an AI agent.

System requirements

OS: macOS 11+, Windows 10+, Linux (Ubuntu 20.04+, Debian 11+, Fedora 35+)
RAM: 50–300 MB (model memory depends on provider)
Disk: ~250 MB

Building from Source

# Prerequisites
# - JDK 25+   (Kotlin Multiplatform)
# - Git

git clone https://github.com/askimo-ai/askimo.git
cd askimo

# Run the app directly
./gradlew :desktop:run

# Create native installers
./gradlew :desktop:package

The project is organized as a Kotlin Multiplatform Compose desktop app with shared core logic in shared-jvm (providers, RAG, plans, skills, agent runtimes) and UI code in desktop/.

Who Might Benefit?

Knowledge workers who want a single, fast desktop client for chatting with any LLM and performing RAG over their own documents.
Developers looking to run code‑generation agents or execute scripts without leaving the chat interface.
Privacy‑conscious users who need local indexing and offline speech capabilities.
Teams that want a consistent skill definition that works across multiple agents.

License & Community

License: AGPL‑v3 (source‑available, requires sharing modifications).
Contributions: Welcome via pull requests; DCO signing required. Localization is handled on Crowdin, so non‑code contributors can help translate.
Support: Discussions forum, documentation site, and a growing star history.

In a nutshell: Askimo bundles chat, RAG, scripting, and autonomous AI agents into a performant, privacy‑first desktop app that works with virtually any LLM you can access.