Back to skill
Skillv1.0.0
ClawScan security
Best Minds · ClawHub's context-aware review of the artifact, metadata, and declared behavior.
Scanner verdict
BenignApr 5, 2026, 11:44 AM
- Verdict
- benign
- Confidence
- high
- Model
- gpt-5-mini
- Summary
- Instruction-only thought‑experiment skill that asks the model to simulate top experts; it requests no credentials, installs, or system access and is internally coherent.
- Guidance
- This skill is instruction-only and technically low-risk: it does not install software or request secrets. Main practical risks are model behavior: it may confidently invent or misattribute quotations and could produce content that looks like a real person speaking (ethical/legal concerns). Before relying on outputs, verify any quoted passages and ask the agent to cite verifiable sources or URLs. If you want stronger safety, restrict the skill from simulating living people without consent or require explicit source citations in every answer.
Review Dimensions
- Purpose & Capability
- okThe name/description (simulate the world's best minds) matches the SKILL.md instructions (choose/top-up experts, base simulations on public statements, quote original words). No unrelated binaries, env vars, or config paths are requested.
- Instruction Scope
- noteSKILL.md contains only high-level runtime instructions: pick the most authoritative person(s), ground answers in that person's public work, and quote them when possible. It does not instruct reading files, accessing credentials, or external endpoints. However, the guidance is open‑ended and can lead models to hallucinate attributions or invented quotes if sources are not verified.
- Install Mechanism
- okNo install spec and no code files — lowest technical risk; nothing will be written to disk or downloaded during install.
- Credentials
- okNo environment variables, credentials, or config paths are requested — access level is minimal and appropriate for the stated purpose.
- Persistence & Privilege
- okalways:false and default model invocation settings (agent may invoke autonomously) — standard for skills. The skill does not request permanent presence or system-wide configuration changes.
