autoresearch

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

• Third-party content exposure detected (medium risk: 0.30). The required runtime workflow reads first-party repository files at paths chosen from user inputs (e.g., program.md and /tmp/autoresearch-<tag>-<timestamp>.md, plus <repo>/results.tsv) and then passes their contents/ledger text to subagents for hypothesis selection, so outsider-authored text can be ingested via the target repo.