bailian-gen

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

• Third-party content exposure detected (medium risk: 0.30). The runtime workflow ingests outsider-authored free text directly via user-supplied prompt/message fields (e.g., bl image generate --prompt, bl omni --message, bl video generate --prompt, bl video edit --prompt) that the CLI sends to the model.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly requires reading the shared protocol at ../bailian-protocol/SKILL.md before executing, meaning runtime behavior and prompts are governed by that external file (../bailian-protocol/SKILL.md) and the skill will stop or request installation if it is missing.