deep-research

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

• Third-party content exposure detected (high risk: 0.85). Phase 2 bibliography_agent ingests outsider-authored free text from public web content at runtime via systematic literature search (e.g., database/web results and scraped abstracts/full-text), which is then placed into the agent’s LLM context for screening/annotation.

MEDIUM W021: Hidden or invisible Unicode characters detected (potential obfuscation or prompt injection).

• Hidden Unicode characters detected (1 type(s) found)