Ernest Hysa cb6f6b65ea fix(research): validate session_id to block path traversal ...

Every research endpoint interpolates session_id into filesystem paths
(Path('data/deep_research') / f'{session_id}.json') without checking
for traversal sequences. A crafted ID like '../../data/auth' reaches
arbitrary JSON files — readable via research_detail (which also leaks
file paths in error messages), writable via research_archive, and
deletable via research_delete.

Add _validate_session_id() which rejects anything outside
[a-zA-Z0-9-]{1,128}. Called before filesystem access in all 12
endpoints that accept a session_id path parameter.

2026-06-01 23:25:38 +01:00

1.8 KiB

Raw Permalink Blame History

View Raw