gpt-image-prompt-lib

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly pulls and syncs public GitHub prompt repositories via scripts/sync_cases.py (sources listed in README and SKILL.md), populates references/case-index.md and references/case-library.md with community-authored prompts, and the workflow (SKILL.md Steps 2, 5–7) requires reading those untrusted, user-generated prompts and copying their style/instructions into a synthesized prompt that is then passed to the image_gen tool—so third-party content can directly influence tool actions.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill runs scripts at runtime (check_updates.py / sync_cases.py) that pull case data from upstream git repositories (e.g., git@github.com:EvoLinkAI/awesome-gpt-image-2-prompts.git and git@github.com:Anil-matcha/Awesome-GPT-Image-2-API-Prompts.git), and those fetched prompt cases are read and injected verbatim/near-verbatim into synthesized prompts that directly control the agent's image_gen instructions, so these remote repos are runtime dependencies that can control prompts.