Skill Maker (ChenXi)
v1.2.0用于根据对话记忆提炼可复用工作流并生成高质量 SKILL.md，适用于“从聊天沉淀方法论”“把一次性操作升级为可复用技能”场景。
⭐ 3· 493·2 current·2 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
Name/description match the behavior: it is an instruction-only 'skill-maker' that explains how to extract workflows from conversations and project files and produce SKILL.md. It does not request unrelated binaries, credentials, or installs.
Instruction Scope
The instructions explicitly tell the agent to automatically retrieve context from recent conversations, related project files, and historical skills drafts — this is coherent with the purpose but means the agent will access potentially sensitive chat logs and files. The skill also prescribes writing outputs to ~/.openclaw/skills and forbids ~/.../workspace/skills. It instructs de-identification of names/keys/paths, which mitigates but does not eliminate the privacy surface; users should confirm what conversation/file access the agent is allowed.
Install Mechanism
No install spec or code is included (instruction-only), so nothing will be downloaded or written by an installer step.
Credentials
The skill declares no required environment variables, credentials, or config paths — consistent with an instruction-only authoring tool.
Persistence & Privilege
always is false and the skill does not request elevated or system-wide privileges. It expects to write generated SKILL.md files into the user's ~/.openclaw/skills directory (normal for this purpose) and does not instruct modifying other skills' configs.
Assessment
This skill is coherent for generating SKILL.md from conversation and project context, but it will (by design) access recent chat history and project files to assemble context. Before enabling or running it: 1) confirm the agent has permission to read the chat logs and files it will scan and remove any sensitive sources you don't want included; 2) review the generated SKILL.md for residual secrets or identifiers despite the '去敏' step; 3) note the output path (~/.openclaw/skills) and move or inspect files as needed; and 4) only run any suggested 'clawhub publish' or external publish steps if you intend to make the skill public.Like a lobster shell, security has layers — review code before you run it.
latestvk97bkyh6y462p4d43fzrxh79rx82abrq
License
MIT-0
Free to use, modify, and redistribute. No attribution required.