首页 / 技能 / Self Improving Agent — 自我优化智能体

📦 Self Improving Agent — 自我优化智能体

v1.0.2

本地运行的智能体系统，可分析对话质量、发现改进点并持续优化回复策略，无需网络或外部凭据。

0· 25·1 当前·1 累计
handy01 头像by @handy01·MIT
智能体自动化数据分析生产力工具开发工具
下载技能包
License
MIT
最后更新
2026/4/21
0
安全扫描
VirusTotal
无害
查看报告
OpenClaw
安全
high confidence
该技能的代码与运行指令符合本地自我改进/日志辅助工具特征：仅在智能体工作区内读写文件，无网络调用或外部凭据，无需安装依赖。
评估建议
This skill appears coherent and local: before installing, confirm you are comfortable with it creating and appending to files in the agent workspace (default /root/.openclaw/workspace/improvement_log.md and SOUL.md). Note that "automatic analysis" requires the host agent to call analyze_conversation or otherwise integrate the skill — the included code does not autonomously monitor conversation streams. There are no network calls or credential requirements in the code, but if you plan to use it i...
详细分析 ▾
ℹ 用途与能力
Name/description match the included Python implementation: analyzing conversations, logging improvements, suggesting SOUL.md updates, and generating reports. No unrelated credentials, binaries, or installs are requested. Note: SKILL.md states the skill "automatically analyzes conversations after each session" but the Python code provides methods (analyze_conversation) and a CLI — it does not itself hook into a session stream or run a background monitor. Integration is required to achieve automatic analysis.
ℹ 指令范围
SKILL.md instructions and the code stay within the stated scope: analyzing text, appending to an improvement_log.md, reading/writing SOUL.md, and producing reports. The skill writes to and reads from files under a workspace (default /root/.openclaw/workspace). This file I/O is expected for its purpose but means conversation-derived insights and manual logs will be stored on disk in that workspace.
✓ 安装机制
No install spec is provided (instruction-only with an included code file). No external downloads, package installs, or archive extraction occur — lowest install risk.
✓ 凭证需求
The skill requests no environment variables or credentials. It uses a default workspace path; no access to unrelated secrets or system config is requested.
✓ 持久化与权限
always is false and the skill does not request permanent platform-wide privileges. It reads/writes files only within its workspace and does not modify other skills or global agent configuration.
安全有层次，运行前请审查代码。

License

MIT

可自由使用、修改和再分发，需保留版权声明。

运行时依赖

无特殊依赖

版本

latestv1.0.22026/4/20

- 内部元数据（_meta.json）已更新，用户可见文档与功能未变动。 - 版本号与 SKILL.md 内容保持不变。

● 无害

安装命令

点击复制
官方npx clawhub@latest install handy01-self-improving-agent
镜像加速npx clawhub@latest install handy01-self-improving-agent --registry https://cn.longxiaskill.com

技能文档

# 自我改进智能体 面向 OpenClaw 智能体的智能自我改进系统，可分析对话质量并持续优化表现。 ## 功能 - 📊 质量分析：评估对话效果 - 🎯 改进追踪：发现可提升之处 - 📝 学习日志：记录洞察与教训 - 📈 周报生成：汇总改进情况 - 🔄 策略优化：随时间调整回应模式 ## 安装 ``bash clawhub install self-improving-agent ` ## 用法 ### 自动分析 每次对话结束后，该 skill 会自动进行分析。 ### 手动记录改进 `python from self_improving import SelfImprovingAgent sia = SelfImprovingAgent() sia.log_improvement("Need to be more concise in technical explanations") ` ### 生成周报 `python report = sia.generate_weekly_report() print(report) ` ## 配置 | 设置 | 默认值 | 说明 | |---------|---------|-------------| | improvement_log | ./improvement_log.md | 学习日志文件 | | soul_file | ./SOUL.md | 人格锚定文件 | | auto_analyze | true | 自动分析对话 | ## 集成 最佳搭配： - memory-manager`：用于追踪改进历史 - 自定义智能体人格（SOUL.md） ## 许可证 MIT 许可证 —— 帮助智能体日日精进！

数据来源：ClawHub ↗ · 中文优化：龙虾技能库