lazyweb-design-best-practices

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.70). Almost all links are raw GitHub markdown or GitHub repos (low-risk to fetch as text), but the presence of an external install script (https://www.lazyweb.com/install.sh) and an explicit curl | bash instruction in the skill prompt is a high-risk executable-download/execute vector that could distribute malware.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.95). The required runtime workflow fetches and reads outsider-authored free text from public GitHub raw URLs (e.g., WebFetch/lazyweb_fetch_best_practice retrieving other repos’ SKILL.md content), which is then ingested into the agent’s LLM context.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 1.00). The file explicitly instructs the agent to fetch and apply remote SKILL.md content at runtime (e.g. https://raw.githubusercontent.com/emilkowalski/skill/main/skills/emil-design-eng/SKILL.md) and even suggests running a remote installer via curl -fsSL https://www.lazyweb.com/install.sh | bash, so external content would directly control agent instructions and could execute remote code.