linkedin

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

• Third-party content exposure detected (high risk: 0.85). Runtime path: the skill calls UnifAPI MCP operations like linkedin/companies/{slug}, linkedin/companies/{slug}/people, linkedin/users/{username}/posts, and .../comments, which return public LinkedIn text (e.g., post bodies/comments) authored by third parties; that free text is then placed into the agent’s LLM context as response data.