blip-2-vision-language

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

• Third-party content exposure detected (high risk: 0.80). The skill includes code that fetches arbitrary web images (e.g., references/troubleshooting.md "Handle URL images" with requests.get(url)) and then feeds those untrusted, third-party images into the BLIP-2 model and agent workflows (including LangChain/agent integration), so external content can be interpreted and materially influence model outputs and subsequent actions.