← Back to Skills Marketplace
kingofzhao

Skill Collision Engine

by KingOfZhao · GitHub ↗ · v1.0.0 · MIT-0
cross-platform ⚠ suspicious
82
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install skill-collision-engine
Description
Skill碰撞引擎 —— 让Skill之间自动四向碰撞生成新Skill的元引擎,Skill工厂的自进化核心
Usage Guidance
Do not install blindly. Key points to consider before proceeding: - The skill ships only documentation (no code). The SKILL.md advertises a Python API and install commands but no Python module is present — ask the author for the actual implementation or a link to the exact code that will run. - The runtime instructions ask the agent to read verification logs across all Skills and to fetch from many external sources; verify what files and external endpoints it will access and whether those contain sensitive data. - The skill does not declare any API keys or credentials even though some sources (YouTube, GitHub with high rate) may require them; demand explicit configuration guidance and least-privilege requirements. - If you consider using it, run it in a restricted sandbox first (no access to private Skills, secrets, or system directories), require explicit consent for scanning other Skills, and audit any produced outputs and network calls. - If you cannot obtain the referenced implementation (the SkillCollisionEngine module) or a clear security/configuration model from the author, treat this skill as untrusted and avoid granting it broad file or network access.
Capability Analysis
Type: OpenClaw Skill Name: skill-collision-engine Version: 1.0.0 The skill-collision-engine is a meta-framework designed to automate the creation and evolution of other skills by analyzing and merging cognitive frameworks. The logic described in SKILL.md and HEARTBEAT.md focuses on file-based analysis of known/unknown data sets and the generation of new skill definitions. There is no evidence of data exfiltration, malicious execution, or harmful prompt injection; the instructions are entirely aligned with the stated purpose of a 'skill factory' and use standard OpenClaw patterns for skill management.
Capability Assessment
Purpose & Capability
The stated purpose (a meta-engine that collides Skills and self-evolves) is coherent with the described operations (reading other Skills, external knowledge sources, producing new Skills). However the SKILL.md includes a Python API (SkillCollisionEngine class and methods) and install commands while the package contains no code files — this is an inconsistency between claimed capability and delivered artifacts.
Instruction Scope
Instructions direct the agent to read verification logs across 'all published Skill' (VERIFICATION_LOG.md), pull from multiple external sources (ArXiv, GitHub, YouTube, web/PDF, human dialog), and run periodic 'heartbeat' tasks. Those actions imply broad filesystem and network access and potential aggregation of sensitive or private Skill data; yet the skill does not declare limits or required permissions. The SKILL.md also gives executable examples (Python import) that do not exist in the package.
Install Mechanism
This is an instruction-only skill with no install spec and no bundled binaries — lowest install risk. The README suggests 'clawhub install' or copying files, but nothing will write or execute shipped code because no code files are present.
Credentials
No environment variables or credentials are declared even though the SKILL.md describes accessing external services (GitHub, YouTube transcripts) and reading many local verification logs. Either the skill expects to run with existing agent permissions (which gives it broad access) or it omits declaring required API keys — both are notable omissions that reduce transparency.
Persistence & Privilege
The skill is not set to always:true and uses normal autonomous invocation defaults. However the documentation prescribes recurring 'heartbeat' and daily auto-collision tasks and self_evolve routines that would run periodically and read/aggregate data; without shipped code or explicit scheduler hooks it's unclear how/when these run. If the agent executes these autonomously with broad file/network access, the blast radius increases.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install skill-collision-engine
  3. After installation, invoke the skill by name or use /skill-collision-engine
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
Skill工厂第3个: Skill-Skill认知碰撞+SOUL领域适配+工厂元进化+碰撞矩阵
Metadata
Slug skill-collision-engine
Version 1.0.0
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 1
Frequently Asked Questions

What is Skill Collision Engine?

Skill碰撞引擎 —— 让Skill之间自动四向碰撞生成新Skill的元引擎,Skill工厂的自进化核心. It is an AI Agent Skill for Claude Code / OpenClaw, with 82 downloads so far.

How do I install Skill Collision Engine?

Run "/install skill-collision-engine" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is Skill Collision Engine free?

Yes, Skill Collision Engine is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does Skill Collision Engine support?

Skill Collision Engine is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created Skill Collision Engine?

It is built and maintained by KingOfZhao (@kingofzhao); the current version is v1.0.0.

💬 Comments