N.E.K.O. is an open-source AI companion platform that runs locally with real-time perception, memory, and multi-form avatars. It supports Docker deployment, optional web UI, and plugin/agent ecosystems.
Collecting history — the radar snapshots this repo daily. The trend line appears after 3 days of data (1 so far).
What it is
N.E.K.O. is an open-source AI companion platform designed to run locally. It emphasizes real-time perception (voice and vision), multi-dimensional memory (five memory types), and multiple avatar forms (Live2D/VRM/MMD/PNGTuber/cat desktop pet). It includes an Agent execution capability for interacting with browsers and apps, and a plugin ecosystem with 14+ AI service providers.
How it works
The core architecture is modular, with components for memory (facts, reflection, persona), main servers (memory_server, main_server), an agent subsystem, front-end UIs (React chat and plugin manager), and a plugin framework. It supports real-time API clients and offline text APIs, multiple model/providers, and a cross-server communication layer. It can be deployed via Windows/macOS/Linux one-click package or via Docker (compose or run).
Getting started
- One-click package: Windows/macOS/Linux, run N.E.K.O.exe, N.E.K.O.app, or n.e.k.o. (macOS requires Gatekeeper adjustments).
- Docker deployment (Linux): two main options are Docker Compose and Docker Run, with environment variable configuration and SSL certificate handling. See example docker-compose.yml and docker run commands in the README. Health checks and logs are provided.
- Local web UI paths include api_key, character_card_manager, memory_browser for configuration and memory review.
Recent releases
- Nightly Build 2026-08-05 (unsigned nightly) for testing on macOS.
- v0.8.3 (2026-06-27): introduces 神志凝神模式 (focus mode) and cat-mode enhancements, with new visuals and memory-related improvements.
- v0.8.2 (2026-06-12): desktop chat stabilization, UI improvements, yarn-ball minimized mode, multi-display placeme.
- v0.8.1 (2026-05-18): memory system upgrades, smarter proactive conversations, new TTS providers (ElevenLabs, Groq) and presets.
- v0.8.0 (2026-05-08): YUI introduction, new onboarding guides, long-term memory upgrade, and perception-related updates.
Traction
- 2376 stars, 269 forks, 82 open issues as of the provided data.
Behind the repo
- The project is Apache-2.0 licensed and emphasizes open-source core drivers with Steam workshop integration and cross-platform support. It provides Steam Creative Workshop support and cross-platform distribution plans.
Caveats
- Licenses: Apache-2.0. Requires Python 3.11 for development and a Node.js (>=20.19) for frontend builds. Nightly unsigned builds exist for testing. SSL setup is documented with multiple configuration options and caveats about certificate handling.
- Large repository history: recommended partial clone for development to avoid downloading obsolete binary resources. Subsections include detailed deployment, environment variables, and SSL certificate requirements.