search-jay-llm-wiki

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

• Third-party content exposure detected (high risk: 0.70). The required workflow invokes $llm-wiki to retrieve and read Wiki page bodies from the fixed project root (wiki/index.md and wiki/ pages), which are outsider-authored free text relative to the operating user’s task context (i.e., not authored by the operating user at runtime), and that retrieved prose is then fed into the agent’s LLM context for background checking.