search

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

• Third-party content exposure detected (medium risk: 0.30). In the required runtime workflow, an outsider can supply free text indirectly via the untrusted “stdin-results” JSON rows (or via a user-supplied URL for --verify), which the agent normalizes and then fetches/reads at Tier 2 evidence time (agbrowse search ... --stdin-results → adaptive-fetch reads page text).