← 返回 Skills 市场
256
总下载
0
收藏
15
当前安装
1
版本数
在 OpenClaw 中安装
/install self-improving-agent-3-0-0
功能描述
Captures learnings, errors, and corrections to enable continuous improvement. Use when: (1) A command or operation fails unexpectedly, (2) User corrects Clau...
安全使用建议
This skill appears to do what it says: provide reminders, capture errors/corrections to .learnings/, and help promote repeatable learnings. Before installing or enabling hooks globally, review the scripts (activator.sh, error-detector.sh, extract-skill.sh) yourself and consider: (1) prefer project-level installation over user-wide (~/.openclaw) if you want to limit injection to a single project, (2) enabling hooks will make the reminder/error detector run in sessions with the agent's permissions—they are lightweight and only emit text or write local files when you run extract-skill.sh, but confirm you trust the skill source, and (3) no credentials are requested, but be cautious about promoting any captured content to shared workspace files if it could include sensitive data. If you want additional assurance, run the scripts in a sandboxed/project directory first and inspect created files before copying them into your global OpenClaw folders.
功能分析
Type: OpenClaw Skill
Name: self-improving-agent-3-0-0
Version: 1.0.0
The 'self-improving-agent' skill bundle is a legitimate development tool designed to facilitate continuous learning for AI agents by logging errors, corrections, and insights into markdown files. It includes utility scripts (e.g., `extract-skill.sh`, `error-detector.sh`) and OpenClaw hooks (`handler.js`) that inject reminders into the agent's context to encourage the capture of learnings. The scripts perform standard file system operations consistent with their stated purpose and include basic security measures, such as path traversal checks in the skill extraction script. No evidence of malicious intent, data exfiltration, or unauthorized execution was found.
能力评估
Purpose & Capability
The name/description (self-improvement / log learnings) match the included files: a bootstrap hook, lightweight reminder and error-detection scripts, templates, and a skill-extraction helper. All artifacts are directly related to capturing and promoting learnings and integrating with OpenClaw/Claude-style hooks.
Instruction Scope
SKILL.md focuses on creating .learnings/ logs, copying hooks into OpenClaw (~/.openclaw/hooks), and enabling optional scripts (activator, error-detector). This is within scope, but installing at user/global level will cause the workspace files and hook behavior to be injected into every session — the user should be aware this promotes reminders across sessions. The error-detector reads CLAUDE_TOOL_OUTPUT (expected), and extract-skill.sh writes files locally when run (expected).
Install Mechanism
No automatic install or remote downloads are declared. The skill is instruction-only with local scripts and hooks. All scripts operate on local files and use standard shell utilities; no network fetches or obscure installers are present.
Credentials
The skill declares no required environment variables or credentials. The only environment usage is reading CLAUDE_TOOL_OUTPUT in the error-detector (documented and relevant). No secrets or unrelated service credentials are requested.
Persistence & Privilege
always:false (normal). The hook/installation recommendations encourage placing files under ~/.openclaw and enabling a hook; that gives the skill workspace-wide presence if the user opts in. Hooks run with the same permissions as the agent, so enabling them at the user/global level affects all sessions — this is expected but worth noting before global enablement.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install self-improving-agent-3-0-0 - 安装完成后，直接呼叫该 Skill 的名称或使用
/self-improving-agent-3-0-0触发 - 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v1.0.0
Initial release of the self-improvement skill for OpenClaw and other agent platforms.
- Enables structured logging of learnings, errors, and feature requests to Markdown files for continuous improvement.
- Provides workflow for promoting important learnings to project memory files (`CLAUDE.md`, `AGENTS.md`, etc.).
- Includes detailed templates and quick reference guides for different types of entries (learnings, errors, feature requests).
- Supports OpenClaw-specific integration, workspace structure, and optional automatic reminder hook.
- Offers generic setup instructions for use in non-OpenClaw agent environments.
元数据
常见问题
Self Improving Agent 3.0.0 是什么？
Captures learnings, errors, and corrections to enable continuous improvement. Use when: (1) A command or operation fails unexpectedly, (2) User corrects Clau... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 256 次。
如何安装 Self Improving Agent 3.0.0？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install self-improving-agent-3-0-0」即可一键安装，无需额外配置。
Self Improving Agent 3.0.0 是免费的吗？
是的，Self Improving Agent 3.0.0 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Self Improving Agent 3.0.0 支持哪些平台？
Self Improving Agent 3.0.0 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 Self Improving Agent 3.0.0？
由 991200448（@991200448）开发并维护，当前版本 v1.0.0。
推荐 Skills