mirror of
https://github.com/unclecode/crawl4ai.git
synced 2026-06-10 15:58:15 +00:00
Reported by secsys_codex (2026-04-18): /md, /crawl, /llm endpoints pass user URLs to crawler.arun() with no private IP validation. - Add validate_url_destination() to utils.py with opt-out via CRAWL4AI_ALLOW_INTERNAL_URLS=true env var for users who need to crawl internal services. - Integrate into validate_url_scheme() (covers all server.py endpoints). - Add validation at all 4 URL entry points in api.py (handle_llm_qa, handle_markdown_request, create_new_task, handle_crawl_request). - raw: URLs bypass check (inline HTML, no network fetch). - 16 adversarial + source coverage tests added. - secsys_codex added to SECURITY-CREDITS.md. DO NOT PUSH until release day.