blog-post

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

• Third-party content exposure detected (high risk: 0.75). The required workflow delegates “Research [TOPIC]” to a researcher subagent, which (by definition of the skill’s research step) is likely to fetch and summarize public web content at runtime and then write it to research/[slug].md for the main agent to read—i.e., outsider-authored free text enters the LLM context via the saved findings file.