Project N.E.K.O.Developer Documentation

A proactive, multimodal AI companion with optional screen-aware interaction, persistent memory, agent channels, and embodied avatars.

Get Started

Steam Workshop & Community

Available on Steam with Workshop support for sharing character cards, supported avatar assets, previews, and optional reference voice samples.

View on Steam

🎙️

Omni-Modal Dialogue

Voice, text, and vision in a unified conversation loop. Real-time speech with RNNoise neural denoising, AGC, and VAD for ultra-low-latency interaction.

Learn more

💬

Proactive Chat

Optional proactive interaction can use screen context, supported feeds, music, and memes when the corresponding features are enabled. Privacy mode can stop proactive screen viewing.

Learn more

🧠

Five-Dimensional Memory

Per-character working, recent, fact, reflection, and persona layers. BM25 recall works without embeddings; optional local embeddings can improve semantic retrieval.

How it works

🤖

Agent Framework

Optional background tasks through enabled and ready Computer Use, Browser Use, user-plugin, OpenClaw, and OpenFang channels. Individual tasks and all active tasks can be cancelled.

Explore agents

🔌

Plugin Ecosystem

Plugin SDK and marketplace for custom extensions, with a decorator-based API, async lifecycle hooks, inter-plugin messaging, and Agent-facing entries when enabled.

Build a plugin

🎭

Live2D, VRM, MMD & PNGTuber

Four supported avatar formats can run in the main UI and desktop-pet host mode, with format-specific expressions, lip sync, animations, and interaction. Voice registration supports multiple cloud and local backends with provider-specific requirements.

Frontend guide

🌐

Configurable AI Providers & i18n

Multiple core, assist, speech, and related provider profiles are configurable. Provider availability changes by version and region; the product UI and prompts support 8 languages.

Provider list