← 返回 Skills 市场
372
总下载
0
收藏
2
当前安装
1
版本数
在 OpenClaw 中安装
/install self-evolution-engine
功能描述
Autonomous self-improvement engine that learns from interactions, identifies patterns, and evolves behavior over time. Use when: (1) Analyzing interaction pa...
安全使用建议
This skill will read and analyze files in a workspace under your home directory (e.g., .learnings/, memory/, MEMORY.md) and can write changes to agent behavioral files (SOUL.md, AGENTS.md, TOOLS.md). The registry metadata did not declare those config paths or filesystem access. Before installing or running it: (1) Review the full evolution.py script (the provided file was truncated in the package) to confirm there are no network calls or hidden endpoints that could exfiltrate data. (2) Back up SOUL.md/AGENTS.md/TOOLS.md and any memory files so you can revert changes. (3) Run the tool in a sandboxed environment or container first and exercise only read/analysis modes (dry-run) to verify outputs. (4) Prefer disabling autonomous invocation until you confirm behavior and/or add explicit safeguards (approval steps) around any --integrate actions. (5) If you proceed, limit the filesystem workspace it can access or move a copy of your workspace into an isolated test area. If you want higher confidence, provide the remaining portion of evolution.py so it can be fully audited for network I/O, subprocess execution, or other risky operations.
功能分析
Type: OpenClaw Skill
Name: self-evolution-engine
Version: 1.0.0
The self-evolution-engine is a tool designed for autonomous agent self-improvement by analyzing interaction logs and patterns. The core logic in scripts/evolution.py is limited to reading local workspace files (e.g., .learnings/ and memory/), parsing them for recurring errors or successes, and generating markdown reports. While SKILL.md describes high-privilege actions like updating the agent's core behavioral files (SOUL.md, AGENTS.md), the provided Python code does not implement any file-writing or 'integration' logic, and it emphasizes safety measures such as backups, validation, and human oversight.
能力评估
Purpose & Capability
Name/description describe a self-improvement engine and the included script implements pattern detection, rule extraction, validation, and integration into behavioral files (SOUL.md, AGENTS.md, TOOLS.md). That capability broadly matches the stated purpose. However, the registry metadata declares no required config paths or credentials while the skill clearly reads/writes user workspace files under the user's home directory (~/.openclaw/workspace). The lack of declared filesystem/config access in metadata is an inconsistency.
Instruction Scope
SKILL.md instructs the agent to read .learnings/, memory/YYYY-MM-DD.md, MEMORY.md, and session transcripts and to run the provided script to analyze, validate, and integrate behavioral changes. These instructions allow reading potentially sensitive conversation history and writing/updating agent behavior files (SOUL.md, AGENTS.md, TOOLS.md). The skill therefore has the ability to collect broad personal and interaction data and to persist changes to agent behavior — this is a significant scope of access that is not narrowly constrained. The SKILL.md uses a {baseDir} placeholder while the script uses a hard-coded WORKSPACE path under the user's home, which is a mismatch that could cause unexpected behavior.
Install Mechanism
No install spec is present (instruction-only plus included script). This is lowest install risk: nothing is downloaded during install. The provided script will run if invoked, and files will be created under the user's workspace, but there is no external installer or remote download indicated.
Credentials
The skill declares no required environment variables or config paths but both SKILL.md and the script operate on a workspace under the user's home and expect files such as .learnings/, memory/, MEMORY.md, and behavioral files. That means it requires unfettered filesystem access to user data despite metadata not declaring this need. While no external credentials are requested, the requested implicit access to conversational memory and session transcripts is high-impact relative to what the metadata advertises.
Persistence & Privilege
The skill can create an evolution directory and write changes to behavioral files (SOUL.md, AGENTS.md, TOOLS.md) — persistent modifications to agent behavior. 'always' is false, and autonomous invocation is allowed (platform default), but the combination of autonomous invocation plus the ability to modify persisted behavior files increases blast radius. The skill claims validation steps but also offers an auto-integrate flow; metadata does not indicate it will modify persistent config which is an incoherence.
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-evolution-engine - 安装完成后，直接呼叫该 Skill 的名称或使用
/self-evolution-engine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
Autonomous self-improvement engine that learns from interactions, identifies patterns, and evolves behavior over time.
元数据
常见问题
Self Evolution Engine 是什么？
Autonomous self-improvement engine that learns from interactions, identifies patterns, and evolves behavior over time. Use when: (1) Analyzing interaction pa... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 372 次。
如何安装 Self Evolution Engine？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install self-evolution-engine」即可一键安装，无需额外配置。
Self Evolution Engine 是免费的吗？
是的，Self Evolution Engine 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Self Evolution Engine 支持哪些平台？
Self Evolution Engine 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 Evolution Engine？
由 shenmeng（@shenmeng）开发并维护，当前版本 v1.0.0。
推荐 Skills