mateo-berri 9b7a693532 fix(anthropic): read a later system message in place and count only the marks each provider receives ... Claude hoists only the leading run of system messages, so a later one is now ordered where it sits when settling ttls and trimming under drop_params. Before, a breakpoint ahead of a later system message's 1h mark went out as 5m and Claude answered 400. A message-level breakpoint on a turn with no string content is no longer counted, since Claude never receives it. The cache-control hook counts a prompt_cache_breakpoint on a tool call only for Claude targets, so gpt-5.6 keeps the injection point it had before.		2026-10-14 20:55:28 +00:00
..
batches	feat(anthropic): workload identity federation and pluggable identity sources (#44448)	2026-10-03 17:08:30 -07:00
chat	fix(anthropic): read a later system message in place and count only the marks each provider receives	2026-10-14 20:55:28 +00:00
files	chore(typing): clear 670 basedpyright Any errors by validating provider payloads at the boundary (#45388)	2026-10-13 15:35:23 -07:00
messages	feat(anthropic): workload identity federation and pluggable identity sources (#44448)	2026-10-03 17:08:30 -07:00
pass_through	feat(anthropic): generate auto mode safeguard_results with a classifier model on the /v1/messages chat bridge (#45600)	2026-10-14 13:09:11 +00:00
__init__.py	test(unit): add package markers to migrated unit test directories	2026-09-20 11:53:14 +00:00
test_anthropic_common_utils.py	fix(anthropic): hold every prompt_cache_options source and Converse cache point model to the Claude rules	2026-10-14 07:27:57 -07:00
test_anthropic_count_tokens_transformation.py	fix(anthropic): count a leading system run through count_tokens' system parameter (#45463)	2026-10-08 22:34:25 -07:00
test_anthropic_files_and_batches.py	feat(anthropic): workload identity federation and pluggable identity sources (#44448)	2026-10-03 17:08:30 -07:00
test_anthropic_output_format_filter.py	test: finish the non-proxy half of tests/test_litellm (#43281)	2026-09-25 22:43:41 -07:00
test_anthropic_prompt_cache_prediction.py	refactor(anthropic): move prompt cache observation lookup out of the proxy hook (#45748)	2026-10-10 19:06:16 +00:00
test_anthropic_prompt_cache_pricing.py	refactor(anthropic): move prompt cache pricing helper out of the proxy package (#45753)	2026-10-10 20:32:34 +00:00
test_anthropic_reasoning_effort.py	refactor(llms): expose public names for private provider helpers (#45037)	2026-10-07 08:49:25 -07:00
test_anthropic_schema_filter.py	test: move tests/test_litellm/llms into tests/unit/llms (#43191)	2026-09-25 12:43:23 -07:00
test_anthropic_structured_output.py	test: move tests/test_litellm/llms into tests/unit/llms (#43191)	2026-09-25 12:43:23 -07:00
test_anthropic_wif.py	fix(anthropic): fail closed when a token-file or inline-token federation credential is missing a rule or organization id (#45178)	2026-10-07 18:08:19 -07:00
test_azure_ai_cache_pricing.py	test: move tests/test_litellm/llms into tests/unit/llms (#43191)	2026-09-25 12:43:23 -07:00
test_cost_calculation_dict_safety.py	feat(anthropic): workload identity federation and pluggable identity sources (#44448)	2026-10-03 17:08:30 -07:00
test_count_tokens_oauth.py	test: make 77 legacy live tests offline in litellm_utils, router_unit and responses dirs (#45298)	2026-10-08 02:37:55 -07:00
test_message_sanitization.py	feat(anthropic): workload identity federation and pluggable identity sources (#44448)	2026-10-03 17:08:30 -07:00