by SleeeepyZhou
flowchart LR
subgraph Agent
direction LR
A[Agent]
subgraph Perception
P2[Perception2D]
P2N[...] --> P2
P3[Perception3D]
P3N[...] --> P3
end
subgraph Memory
M1[AgentMemory] <--> M2[LongMemory]
end
subgraph Planning
S[TaskSystem]
S3[...] <--> S
end
subgraph Action
T[Action]
AgentTools --> T
end
P2 ==> A
P3 ==> A
M1 <==> A
S <==> A
A <==> T
end
subgraph ENV[Environment]
direction TB
PHYS[Physics]
UI[Game UI]
end
ENV --> Perception
T --> ENV
subgraph MultiAgent System
direction TB
subgraph Agent Network
A1[Agent1] <--> C
A2[Agent2] <--> C
A4[...] <--> C
C[AgentCoordinator]
end
subgraph Model
M[Model]
ModelPool --> M
end
M --> A
subgraph Toolbox
direction LR
T0[Tools] --> t[Interpreter]
t --> GMCP
subgraph GMCP
direction LR
MCP
T1[GodotMCP]
T2[...]
end
GMCP --> T0
end
T0 --> AgentTools
end
Agent --> A1
Native support MCP (Model Context Protocol) through MCP-Godot-sdk.
MCP-Godot-sdk by SleeeepyZhou MIT
Godot-AIUtils by SleeeepyZhou Apache-2.0
MatrixCalc by SleeeepyZhou MIT
This project learns from the renowned Multi-Agent framework Camel, and thanks to Guohao Li and Camel-AI team.
The design of the agent learns from Lilian Weng's article LLM Powered Autonomous Agents.
This project is licensed under the MIT License - see the LICENSE file for details.