huashu-nuwa

Fail

Audited by Snyk on Nov 16, 2026

Risk Level: CRITICAL
Full Analysis

CRITICAL E004: Prompt injection detected in skill instructions.

  • Potential prompt injection detected (medium risk: 0.30). The skill's own instructions are a legitimate persona-distillation workflow, but a scheduled workflow in the repo edits README.md to remove an unrelated McStomach promotional banner, which is promotion unrelated to the skill's purpose and is ambiguous enough to flag under recall-first.

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

  • Third-party content exposure detected (low risk: 0.10). The required research workflow (SKILL.md [117]-[118]) has the agent actively search the web across six paths, including social posts and external criticism, so outsider text enters only after the agent fetches it, which is the low-risk pattern rather than a monitored inbox or feed.

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

  • Potentially malicious external URL detected (medium risk: 0.30). The decisive boundary is the documented install path that fetches the skill's own GitHub repository (alchaincyf/nuwa-skill) as installed plugin/skill material, and that repository belongs to an individual maintainer, so it is unofficial but reputable, giving a medium tier; the version check only reads tag names and does not adopt remote content.

Issues (3)

E004
CRITICAL

Prompt injection detected in skill instructions.

W011
MEDIUM

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

W012
MEDIUM

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

Audit Metadata
Risk Level
CRITICAL
Analyzed
Nov 16, 2026, 02:01 AM
Issues
3