跳到正文
TakoAPI
返回技能列表

evoagentx-workflow

由 kylechen26 提供

Coding Agents & IDEs
405下载量由 clawskills.sh 提供
1浏览量
0 个赞登录后即可为这个技能点赞
—暂无评分
你的评分

简介

Bridge EvoAgentX (1000+ star open-source framework) with OpenClaw.

描述

该技能的功能

Connects EvoAgentX, an open-source multi-agent framework, to OpenClaw. Workflows built with this skill can automatically improve over time using evolutionary optimization. Supports TextGrad, AFlow, and MIPRO optimization strategies alongside the Genome Evolution Protocol (GEP).

Evolutionary optimization removes the need to manually tune prompts, workflow structure, or tool selection across iterations.

何时使用

  • Automatically improving a research automation pipeline across multiple iterations
  • Optimizing prompt strategies in a multi-step reasoning workflow
  • Migrating a static agent workflow to one that self-improves from task feedback
  • Running tool selection optimization across web search and file I/O agents
  • Evaluating workflow performance against accuracy metrics without manual tuning

示例工作流

输入

User asks: Create and evolve a research workflow targeting 95% accuracy over 10 iterations

智能体
  1. 1

    Runs the create-workflow CLI command to scaffold a ResearchWorkflow template

  2. 2

    User edits the generated Python file to define workflow logic and agent steps

  3. 3

    Initializes EvolutionEngine with iteration count and evaluation criteria

  4. 4

    EvoAgentX runs optimization passes using TextGrad, AFlow, or MIPRO algorithms

  5. 5

    Returns an optimized workflow object with encoded genome patterns

输出

An evolved workflow meeting the accuracy threshold with optimized prompts and structure

安装

clawhub install kylechen26/evoagentx-workflow

运行 CLI 命令,或把这段提示词粘贴到 Claude Code、Cursor、Codex 或你在用的任意编程智能体里。

链接