Skill Shell
v0.2.2Evaluate external skills before installation and decide whether to install, reject, or absorb only the useful ideas. Use when a user shares a ClawHub/GitHub...
⭐ 0· 418·7 current·7 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, and SKILL.md all describe the same responsibility: reviewing/triaging candidate skills. No unexpected binaries, env vars, or installs are requested — everything is proportionate to a review/meta skill.
Instruction Scope
The instructions are intentionally broad (static review + dynamic validation). They recommend inspecting package contents and, when required, performing dynamic checks which could involve network fetches or running installers. That scope is appropriate for a vetting workflow, but it means dynamic validation steps should only be performed with explicit user approval and in an isolated environment.
Install Mechanism
No install spec and no code files — the skill is instruction-only, which minimizes disk writes and execution risk.
Credentials
No credentials, env vars, or config paths are requested. The SKILL.md warns about reading sensitive locations (e.g. ~/.ssh, ~/.aws) as red flags rather than instructing the agent to read them, which is appropriate.
Persistence & Privilege
The skill is not force-included (always:false) and does not request persistent privileges or to modify other skills or system settings.
Assessment
This skill is a low-risk, coherent checklist for reviewing other skills. Before using it: (1) do dynamic validation only after explicit user approval and inside an isolated environment (sandbox/container) if the review requires running code or fetching external artifacts; (2) never grant the agent credentials or broad filesystem access to perform a review — instead provide a copy of the package or a controlled environment; (3) watch for dynamic validation steps that run installers, curl/wget to unknown endpoints, or request tokens — those require manual oversight; and (4) treat the skill as guidance (absorb ideas) rather than an automatic executor of installation steps unless you trust the source.Like a lobster shell, security has layers — review code before you run it.
latestvk97ey3pqdjpxfhv28qdrmd5y5n82r6j3openclawvk97ey3pqdjpxfhv28qdrmd5y5n82r6j3securityvk97ey3pqdjpxfhv28qdrmd5y5n82r6j3workflowvk97ey3pqdjpxfhv28qdrmd5y5n82r6j3
License
MIT-0
Free to use, modify, and redistribute. No attribution required.