← 返回 Skills 市场
146
总下载
0
收藏
1
当前安装
3
版本数
在 OpenClaw 中安装
/install self-evolving-harness
功能描述
Self-Evolving Harness — 聚焦增强 OpenClaw 的自主性与可控性。从领域意图理解、自主知识沉淀到自主评测与本地 Agent 架构的端到端自主进化，拒绝在 Skill 层面无序熵增。搭配自动化软件研发流水线；反脆弱，即便记忆丢失，Agent 依然有序运行和进化。
安全使用建议
This package looks like a stub or marketing copy rather than a working skill. Before installing or enabling it: (1) treat it as non-functional — it contains no runtime code or instructions; (2) ask the publisher for a full implementation, a README describing runtime steps, and a trustworthy source/homepage; (3) if they provide an implementation later, review the SKILL.md for exact commands, any install URLs, and required environment variables before enabling autonomous invocation; (4) avoid trusting this skill with sensitive credentials or making it always-enabled until its behavior and dependencies are explicit and auditable.
功能分析
Type: OpenClaw Skill
Name: self-evolving-harness
Version: 0.0.3
The skill bundle contains only metadata and a high-level conceptual description in SKILL.md. There is no executable code, scripts, or actionable instructions that could lead to data exfiltration, unauthorized execution, or prompt injection. The content appears to be a placeholder or documentation for an agent autonomy framework.
能力评估
Purpose & Capability
The README/description claims end-to-end autonomous evolution, knowledge consolidation, evaluation, and CI integration, but there are no code files, no install steps, no required binaries, and no environment variables. The requested surface (nothing) does not match the claimed functionality (significant system and network interactions).
Instruction Scope
SKILL.md contains only metadata and a title with no runtime instructions, commands, or guidance for the agent. This makes the skill non-functional as described and prevents verifying what it would actually do at runtime.
Install Mechanism
There is no install spec (instruction-only). That minimizes direct install risk but also means there is no implementation to inspect.
Credentials
The skill requests no credentials or config paths, which superficially reduces risk. However, given the ambitious description, the absence of any required credentials or integrations is unexpected and suggests the package is incomplete or misrepresented.
Persistence & Privilege
always is false and no special persistence or system-wide config changes are requested. The skill is user-invocable only and does not request elevated presence.
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-evolving-harness - 安装完成后，直接呼叫该 Skill 的名称或使用
/self-evolving-harness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
v0.0.3
更新描述至最新项目定位
v0.0.2
补充产品定位描述
v0.0.1
Initial release of Self-Evolving Harness.
- Introduced a self-evolving AI Agent framework focused on controllability and autonomous improvement.
- Defined three core modules: KIVO (cognitive infrastructure), R&D Pipeline (execution infrastructure), and AEO (Agent effect operations).
- Documentation placeholders provided for future development.
元数据
常见问题
Self-Evolving Harness 是什么？
Self-Evolving Harness — 聚焦增强 OpenClaw 的自主性与可控性。从领域意图理解、自主知识沉淀到自主评测与本地 Agent 架构的端到端自主进化，拒绝在 Skill 层面无序熵增。搭配自动化软件研发流水线；反脆弱，即便记忆丢失，Agent 依然有序运行和进化。 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 146 次。
如何安装 Self-Evolving Harness？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install self-evolving-harness」即可一键安装，无需额外配置。
Self-Evolving Harness 是免费的吗？
是的，Self-Evolving Harness 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Self-Evolving Harness 支持哪些平台？
Self-Evolving Harness 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-Evolving Harness？
由 yuchangxu（@yuchangxu1989-openclaw）开发并维护，当前版本 v0.0.3。
推荐 Skills