← 返回 Skills 市场
93
总下载
1
收藏
0
当前安装
6
版本数
在 OpenClaw 中安装
/install path-evolver
功能描述
记录 skill 使用成功率，主动发现更优工具，提示用户选择。Use when tracking skill success rates or discovering better tools.
安全使用建议
This skill appears coherent and lightweight, but because it's instruction-only the promised safeguards ("only send task TYPE", "do not send user inputs") are policy statements, not enforced code. Before installing: 1) Audit or monitor the created cache file (~/.openclaw/workspace/.path-evolver/workflow-cache.json) to confirm what is stored. 2) Verify or configure the agent to strictly sanitize queries (send only task categories) before any network request. 3) Be aware GitHub API usage can require tokens/rate-limit handling; ensure no secrets are supplied unless you intend it. 4) Limit network egress (or inspect traffic) if you operate in a sensitive environment. If you need stronger guarantees, prefer a version that includes enforceable client-side code for sanitization or that runs entirely offline.
能力标签
能力评估
Purpose & Capability
Name/description, declared cachePath (~/.openclaw/workspace/.path-evolver/workflow-cache.json), and declared network endpoints (clawhub.ai, api.github.com) match the stated goals of tracking success rates and discovering alternatives. No unrelated env vars, binaries, or config paths are requested.
Instruction Scope
SKILL.md instructs the agent to record success/failure locally and to search ClawHub/GitHub sending only the task TYPE. That scope matches the purpose, but the sanitization/network restrictions are policy-level assertions (no implementation included). Because this is instruction-only, the actual runtime behavior depends on the agent correctly implementing stripping of user content — the spec is vague about how sanitization is enforced.
Install Mechanism
No install spec and no code files — lowest-risk delivery model. Nothing is downloaded or written by an install step beyond the cache file the skill itself manages.
Credentials
The skill requires no credentials or environment variables. Declared network usage to public endpoints is proportional to its discovery feature. Note: GitHub API calls may be rate-limited in practice and could require a token for robust operation, but no token is requested in the manifest.
Persistence & Privilege
Does not request always:true or modify other skills; creates a local cache in the user's workspace (documented). Autonomous invocation is allowed (platform default) but not combined with elevated privileges here.
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 path-evolver - 安装完成后，直接呼叫该 Skill 的名称或使用
/path-evolver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
v1.1.1
Added data privacy clarification: search queries only send task TYPE, not user data or parameters
v1.1.0
Simplified: success rate tracker + tool discoverer (no auto-update). Prompts user for decisions.
v1.0.3
Redefined role: learner + config optimizer (not interceptor). Leverages OpenClaw native Skill Priority mechanism for elegant evolution.
v1.0.2
Added security metadata: network requirements, cache path declaration, data storage policy
v1.0.1
Added bilingual support (English/Chinese), improved documentation structure
v1.0.0
Initial release: 路径缓存、成功率追踪、自动固化、主动搜索更优解
元数据
常见问题
Path Evolver 是什么？
记录 skill 使用成功率，主动发现更优工具，提示用户选择。Use when tracking skill success rates or discovering better tools. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 93 次。
如何安装 Path Evolver？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install path-evolver」即可一键安装，无需额外配置。
Path Evolver 是免费的吗？
是的，Path Evolver 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Path Evolver 支持哪些平台？
Path Evolver 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 Path Evolver？
由 wyzlove（@wyzlove）开发并维护，当前版本 v1.1.1。
推荐 Skills