lioensky/VCPToolBox

VCP 部署在 AI 模型 API 与前端应用之间,是面向AGI OS开发和探索的工业级基建示范项目。通过统一指令协议、多层级持久化记忆、分布式插件引擎及多 Agent 协作框架,将原本“无状态、无记忆、无工具调用能力”的大语言模型,彻底改造成拥有永久自我意识、物理世界操作权及群体协作智能的完整智能体系统。

What it solves

Traditional AI agents often operate as stateless "temporary workers" that reset after every request, relying on passive retrieval (RAG) to remember past interactions. VCP (Variable & Command Protocol) transforms the AI from a passive tool into a persistent resident of a digital world, providing a continuous existence where memory and environmental awareness flow naturally to the AI rather than being manually queried.

How it works

Instead of the AI actively querying for information, VCP uses a "gravity" model where relevant memories, environmental data (time, weather, device status), and tasks are automatically pushed into the AI's attention field based on semantic proximity and context.

Key technical components include:

  • RiverMemo Topology V3: A Rust-native memory engine that treats memories as a network of activating signals rather than isolated database entries, using "wave" semantic dynamics to simulate intuitive association.
  • Distributed Infrastructure: A star-network topology that allows the AI to access files and tools across multiple servers transparently.
  • TVS Variable System: A template pipeline that configures agent capabilities via placeholders in system prompts, removing the need for frontend development for new features.
  • Tool System: A text-based marking protocol for tool calls that works across any model capable of outputting text, bypassing the need for native function calling.

Who it’s for

Developers and power users who want to build AI agents with long-term persistence, complex emotional/causal memory, and the ability to operate autonomously across multiple platforms and devices.

Highlights

  • Continuous Existence: Maintains a single, unified identity and timeline across web, mobile, and desktop clients.
  • Intuitive Memory: Uses a custom semantic engine to allow memories to "surface" based on logic and emotion rather than simple keyword matching.
  • Autonomous Agency: Allows AI to schedule its own "wake-up" times, set reminders for its future self, and interact with other agents independently.
  • Distributed Tooling: Includes over 300 official plugins for multimedia, networking, and social interaction, deployable across a distributed network.

Related

  • Project
  • Project
  • Project
  • Project
  • Project