Self-hosted, local-first AI assistant server: swappable LLM providers, review-gated memory, an encrypted vault, and wake-word voice. No accounts, no telemetry.
A self-hosted assistant that runs on your machine with the model you choose. Memory is approved by you; keys stay encrypted.
A self-hosted AI assistant server: one web UI, five swappable LLM providers, review-gated persistent memory, an encrypted credential vault, custom wake-word voice, plus capability packs for browser and desktop automation. Packs are a plain-directory format, so writing your own is the intended path. Your vault, memory and files stay on your machine; no accounts, no telemetry, no cloud tenancy.
Recorded from the live app — memory recalled across three tool calls, a real briefing written back.
Where your data lives. Your vault, memory database and pack storage stay on your machine. Model traffic does not: unless you run Ollama, prompts go to the cloud provider you pick. The default voice path is cloud too. edge-tts sends synthesized text to Microsoft; browser speech-to-text sends audio to Google. A fully local pipeline is possible (Ollama for inference plus local voice), but it is the least-tested path: small local models call tools unreliably. Akana's memory/vault tool-calling is unit-tested but not validated end-to-end against a real local model. Treat local inference as experimental. See Security and privacy.
- Quick start - Why Akana - Screenshots - How it works - Providers - Memory - Voice - Capability Packs - Remote access & Telegram - Security and privacy - Configuration - CLI reference - Known limitations - Development - Contributing, license and attribution
From the project README.
Add the radar badge to your README — it shows your project was picked up by MCP Radar and links to this page:
[](https://mcp.liqiwa.com/s/MuhittinYilmazer--akana.html)
Turn any website into a portable, agent-ready Open Knowledge Format (OKF) bundle — no LLM required. Crawl, sync, chat offline, and serve docs to coding agents over MCP.
hritvikgupta/nimbusTalk to your cloud — an open-source AI agent that reads your code, operates AWS & GCP, and opens PRs. Self-hostable.
Ambar-Gupta22/corvusNative C++17 AI agent runtime — local-first, embeddable, MCP-native. Tool-calling, memory, and an async agent loop that runs where Python can't: ROS2, game threads, drones, HFT.
KernelLord/pickysteveSkill router and context picker for Claude Code, Cursor, and Aider. Auto-picks the right skill for every prompt, with prompt-injection filtering for MCP.
Azure-Samples/foundry-hosted-agents-workshopA hands-on, step-by-step workshop for building a Travel Assistant agent on Microsoft Foundry hosted agents with the Agent Framework - growing from a single agent to a multi-agent planner with function
bolivestilo/HomekitHomekit gives any AI agent direct, physical control over Apple Home. Flip lights. Lock doors. Read sensors. Through a single open interface.
The top new MCP servers of the week, every Monday. No spam, unsubscribe anytime.