Commit graph

12 commits

Author SHA1 Message Date
mateo-berri
d3b6ce98d6 fix(responses): encrypt the response id on every streamed event
Background streaming emits event types with no typed model, which arrive
as GenericEvent holding a plain dict. Only typed events had their nested
response id rewritten, so those frames advertised the raw internal id
while their siblings advertised the encrypted one. The raw shape skips
the ownership check, so any other key could retrieve or cancel that
response.

Rewrite the advertised id wherever an event carries one, whichever shape
it arrives in, so a future event type cannot reopen this.
2026-09-03 02:41:04 -07:00
mateo-berri
95f8373e3c test(responses): drive streamed-id regression via production ResponseCompletedEvent shape
The streamed-id regression test built a bare BaseLiteLLMOpenAIResponseObject with a
top-level id, hitting the wrong _encrypt_response_id branch. A real streamed create
emits ResponseCompletedEvent, whose client-visible id lives on event.response.id, so
the test now drives that production event shape and reads collected[0].response.id.
Mutating the alias route gate or disabling the .response.id encryption branch both
fail the test.
2026-08-26 01:52:01 -07:00
mateo-berri
498ba9dd62 fix(proxy): encrypt streamed responses ids on /openai/v1/responses and /responses aliases
The streaming security hook only encrypted response ids when request_route
matched "/v1/responses" exactly, so streamed creates on the /openai/v1/responses
and /responses aliases leaked the plain managed id. A second virtual key could
GET, continue, and DELETE another key's response. Normalize the route (strip the
provider prefix, accept the /responses alias) before gating, mirroring the
non-streaming hook which has no route gate.
2026-08-26 01:38:13 -07:00
Mateo Wang
fab4a9ca26
fix(responses_id_security): decrypt response ids for input_items follow-ups (#32269) 2026-07-06 14:03:15 -07:00
Ishaan Jaffer
e8461b5b97
style: run black formatter on files from main merge 2026-04-17 13:02:59 -07:00
Julio Quinteros Pro
c0ac788709 Fix responses ID security test for new request_cache parameter
The _encrypt_response_id method now receives request_cache=None as a
keyword argument from async_post_call_success_hook. Updated the mock
assertion to expect this parameter.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-04 11:29:51 -03:00
Yuta Saito
2c8ac2c3f1 test: temporarily disable flaky responses_id_security tests 2026-01-13 15:51:37 +09:00
Sameer Kankute
62d860ea7d fix: litellm/tests/test_litellm/test_responses_id_security.py 2026-01-08 17:01:44 +05:30
Ishaan Jaffer
6122ff9fce TestEncryptResponseId 2026-01-07 21:03:22 +05:30
Ishaan Jaffer
b855ad5541 test_encode_decode_helpers_roundtrip_in_cache_context 2026-01-07 18:23:17 +05:30
Ishaan Jaffer
c18f411a5e test_encrypt_response_id_success 2025-11-14 17:28:15 -08:00
Krish Dholakia
72bbdfd3f3
(security) Responses API - prevent User A from retrieving User B's response, if response.id is leaked (#15757)
* feat(responses_id_security.py): encrypt response.id - prevent user A from retrieving user B's response

additional security for retrievals on shared accounts

Closes LIT-1307

* feat(responses_id_security.py): allow admin to disable responses id security check

* test: add initial unit testing

* feat(responses_id_security.py): add streaming support

* docs: document new param

* docs: document new param

* feat(responses_id_security.py): add team id checks - ensure it works for service accounts

prevent service accounts keys from different teams from accessing each other's responses

more secure

* test: add unit testing

* fix: fix linting error
2025-10-25 13:41:59 -07:00