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

VCP (Variable & Command Protocol) addresses the problem of AI agents being "temporary workers"—stateless entities that reset after every request. It replaces the passive query-based model (where AI must actively search for memories or data) with a "gravity" model where relevant information, environment states, and memories naturally flow into the AI's attention field based on semantic relevance and context.

How it works

VCP creates a continuous existence for AI by implementing a following infrastructure:

  • Wave Semantic Dynamics Engine: A Rust-based engine that treats memory as a network of activating signals rather than isolated database entries, allowing for intuitive, associative recall.
  • Semantic Gravity Field: Dynamically calculates what the AI should know at any given moment, folding irrelevant information into summaries and surfacing critical data (time, weather, tasks) before the model even processes the request.
  • Distributed Architecture: A star-topology network that allows for transparent cross-server file access and multi-device synchronization, ensuring the AI is the same entity across all platforms.
  • Tooling System: A text-based marking protocol for tool calls that doesn't rely on native function calling, supporting over 300 plugins for tasks ranging from scientific computing to social interaction.
  • Model Routing: Automatically selects the most appropriate LLM based on the logical depth and topic of the conversation.

Who it’s for

Developers and power users who want to build or deploy AI agents that possess long-term memory, autonomous scheduling (the ability to "wake up" and perform tasks independently), and a consistent identity across multiple interfaces.

Highlights

  • Continuous Identity: Maintains a single, unified timeline of experiences across web, mobile, and desktop clients.
  • Autonomous Living: Allows agents to set their own rhythms, leave notes for their future selves, and enter "flow states" to manage distractions.
  • Associative Memory: Uses a specialized semantic engine to mimic human-like intuition and logical/emotional associations rather than simple vector similarity.
  • High Performance: Implements core retrieval logic in Rust with O(1) lookup times for tens of thousands of tags.