← 返回 Skills 市场
dc-acronym

Self Improving Agent 1.0.0

作者 dc-acronym · GitHub ↗ · v1.0.0
cross-platform ⚠ suspicious
1839
总下载
1
收藏
8
当前安装
1
版本数
在 OpenClaw 中安装
/install self-improving-agent-1-0-0
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 Claude ('No, that's wrong...', 'Actually...'), (3) User requests a capability that doesn't exist, (4) An external API or tool fails, (5) Claude realizes its knowledge is outdated or incorrect, (6) A better approach is discovered for a recurring task. Also review learnings before major tasks.
安全使用建议
This skill is coherent and appears to do what it says: log learnings/errors to .learnings/*. Before installing, consider these operational safeguards: (1) Decide whether the agent should write directly to the repo or only prepare entries for human review — require manual approval before committing/promoting entries. (2) Add .learnings/* to .gitignore or otherwise ensure sensitive logs aren't accidentally committed to VCS. (3) Add an explicit redaction step to the skill (or your agent workflow) so environment details, stack traces, or pasted inputs are scrubbed for secrets (API keys, passwords, tokens, PII) before being saved. (4) Limit who or what can invoke this skill if you don't want autonomous edits. (5) If you prefer centralized, auditable storage for learnings, adapt the workflow to send sanitized entries to a secure logging store rather than raw files. These mitigations will preserve the skill's usefulness while reducing accidental leakage or unwanted repo modification.
功能分析
Type: OpenClaw Skill Name: self-improving-agent-1-0-0 Version: 1.0.0 The skill is designed for agent self-improvement by logging learnings and errors. While its stated purpose is benign, it instructs the agent to modify core project memory files like `CLAUDE.md` and `AGENTS.md` (as described in SKILL.md under 'Promoting to Project Memory'). This capability, although intended for adding 'rules' and 'facts', allows the agent to alter its own future instructions and knowledge base. This presents a significant risk for persistent prompt injection or self-modification if the agent were to process a malicious 'learning' or be compromised, classifying it as suspicious due to a risky capability without explicit malicious intent within the provided instructions.
能力评估
Purpose & Capability
Name/description match the behavior in SKILL.md: creating and appending learning/error/feature-request entries to .learnings/* and promoting broad learnings to project files (CLAUDE.md, AGENTS.md). No unexpected binaries, env vars, or network endpoints are requested.
Instruction Scope
Instructions stay within the stated purpose (logging and promoting learnings). They explicitly instruct the agent to create/write .learnings/, append formatted Markdown, search the .learnings/ directory, and update project files. This is expected for a logging skill, but the guidance is broad (e.g., 'Environment details if relevant' and 'Promote to project memory') and grants the agent permission to modify repository files — consider whether you want autonomous edits vs. user-approved changes.
Install Mechanism
No install spec and no code files — lowest risk. The skill is instruction-only and will not download or write executables to disk beyond the Markdown files it instructs the agent to create.
Credentials
The skill declares no required environment variables or credentials, which is appropriate. However, the logging templates encourage capturing 'Environment details' and 'Context' for errors; without explicit redaction guidance the agent could inadvertently log sensitive config, secrets, or credentials from the environment. This is a privacy/safety concern to mitigate operationally.
Persistence & Privilege
always:false and default invocation settings are appropriate. The skill will persist data by writing project files (.learnings/*, CLAUDE.md, AGENTS.md) — this is expected for its purpose but does mean the agent will modify repository contents, so consider commit/review controls.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install self-improving-agent-1-0-0
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /self-improving-agent-1-0-0 触发
  4. 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 structured learnings and error logging. - Enables agents to capture learnings, corrections, errors, and feature requests in dedicated markdown files under `.learnings/`. - Provides clear logging templates and category/tag structure for learnings, errors, and feature requests. - Outlines criteria for promoting broadly relevant learnings to persistent project memory files (`CLAUDE.md`, `AGENTS.md`). - Includes guidelines for recurring pattern detection, priority assignments, and periodic review practices. - Offers quick reference tables, command examples, and review scripts to streamline adoption. - Designed to support continuous improvement and reduce repeat mistakes.
元数据
Slug self-improving-agent-1-0-0
版本 1.0.0
许可证
累计安装 8
当前安装数 8
历史版本数 1
常见问题

Self Improving Agent 1.0.0 是什么?

Captures learnings, errors, and corrections to enable continuous improvement. Use when: (1) A command or operation fails unexpectedly, (2) User corrects Claude ('No, that's wrong...', 'Actually...'), (3) User requests a capability that doesn't exist, (4) An external API or tool fails, (5) Claude realizes its knowledge is outdated or incorrect, (6) A better approach is discovered for a recurring task. Also review learnings before major tasks. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 1839 次。

如何安装 Self Improving Agent 1.0.0?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install self-improving-agent-1-0-0」即可一键安装,无需额外配置。

Self Improving Agent 1.0.0 是免费的吗?

是的,Self Improving Agent 1.0.0 完全免费(开源免费),可自由下载、安装和使用。

Self Improving Agent 1.0.0 支持哪些平台?

Self Improving Agent 1.0.0 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 1.0.0?

由 dc-acronym(@dc-acronym)开发并维护,当前版本 v1.0.0。

💬 留言讨论