paper-collage-ad-codex

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.80). These GitHub URLs point to an unfamiliar user repository that the skill instructs you to git-clone and run included shell/scripts (including model-download/setup scripts), which is a common vector for distributing malicious code or unsafe binaries.

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

• Third-party content exposure detected (high risk: 0.70). The required workflow ingests user-provided free text into the LLM context via the “Script & Storyboard” step (generate dialogue/scene descriptions from product info and target audience), which is not shown as restricted to only operating-user-authored text.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill instructs the user to git-clone and install remote code from https://github.com/Jane-xiaoer/paper-collage-ad-codex.git and then run scripts from that clone (e.g., setup and various node/bash scripts), meaning fetched remote code is executed as part of the skill runtime.