跳至正文
OpenAgentSkill
浏览已开放 · 目录记录日期 2026-11-21 · 自动更新和账户服务暂时关闭。
开发与测试

AI Agent Core Architectures

Fundamental principles of AI Agent cognitive loops, contrasting ReAct and Plan-and-Solve paradigms.

j4flmao · 已有静态检查记录

安装与使用

运行命令前先阅读来源与依赖说明。复制命令不会自动安装或执行。

npx skills add j4flmao/agent-skills --skill AI Agent Core Architectures

技能说明


name: "AI Agent Core Architectures" description: "Fundamental principles of AI Agent cognitive loops, contrasting ReAct and Plan-and-Solve paradigms."

Core Architectures: The Autonomous Cognitive Loop

The existence of an autonomous agent is defined not by static inference, but by the continuous, recursive execution of the cognitive loop: Perceive -> Think -> Act -> Observe. This loop bridges the gap between latent semantic space and deterministic environment execution.

First Principles of Agentic Flow

Every framework-agnostic architecture reduces to this state machine. The agent's cognition is a sequence of discrete state transitions bounded by token limits and environment feedback.

  1. Perception: Ingestion of environment state. The synthesis of system prompts, historical context, and the immediate state of the world.
  2. Thought (Reasoning): The generation of latent reasoning tokens. This is the derivation of intent, mapping perception to actionable trajectory.
  3. Action: The emission of structured payloads designed to mutate the environment or retrieve novel state.
  4. Observation: The ingestion of the deterministic result of the action, closing the loop.

Architectural Paradigms

ReAct (Reason + Act)

The interleaving of reasoning traces with action execution. ReAct assumes high environmental volatility, requiring continuous recalibration. It sacrifices long-horizon coherence for immediate, localized adaptability.

Plan-and-Solve

The temporal decoupling of strategy from execution. The agent first synthesizes a comprehensive graph of execution steps, then traverses the graph sequentially. Plan-and-Solve assumes low environmental volatility but requires profound foresight. It excels in complex, multi-dependent task resolution but is brittle to unexpected state mutations during execution.

The Necessity of Self-Reflection

Without self-reflection, an agent is an open-loop controller doomed to terminal error spirals. Self-reflection acts as the error-correction mechanism, forcing the agent to evaluate the delta between expected observation and actual observation, dynamically altering its system prompt or execution graph to converge on the goal state.

%%{init: {"theme": "default", "flowchart": {"useMaxWidth": true}}}%%
flowchart TD
    Start([Goal Initialization]) --> Perceive
    Perceive[Perceive Environment State] --> Reflect{Reflection/Evaluation}
    Reflect -- "State aligns with Goal" --> Success([Terminal Success])
    Reflect -- "State divergence" --> Plan[Synthesize Execution Graph]
    Plan --> Think[Reason Next Step]
    Think --> Act[Execute Action Payload]
    Act --> Observe[Observe Environment Feedback]
    Observe --> Perceive