Luis Pater c2dfec28d5 fix(claude): map model_context_window_exceeded to length in chat completions ...

- Map Claude model_context_window_exceeded stop reason to finish_reason length in OpenAI Chat Completions and remove the synthetic context_length_exceeded HTTP 400 error path so generated content, tool calls, and usage are preserved.
- Preserve finish_reason length when non-streaming Chat Completions responses are truncated inside a tool call by max_tokens or model_context_window_exceeded.

Closes: #6861

2026-10-22 17:21:21 +08:00

..

helps

fix(gemini): preserve usage after terminal stream cancellation

2026-10-21 10:33:05 +08:00

testdata/tool_search_disabled

…

aistudio_executor_test.go

…

aistudio_executor.go

…

aistudio_issue6258_terminal_test.go

…

antigravity_caqs_replay_test.go

…

antigravity_connection_trace_test.go

fix(antigravity): align Hub requests and invalidate versioned model caches

2026-10-20 18:46:02 +08:00

antigravity_executor_auth.go

…

antigravity_executor_buildrequest_test.go

fix(antigravity): align Hub requests and invalidate versioned model caches

2026-10-20 18:46:02 +08:00

antigravity_executor_compaction_test.go

…

antigravity_executor_credits_test.go

fix(antigravity): align Hub requests and invalidate versioned model caches

2026-10-20 18:46:02 +08:00

antigravity_executor_credits.go

fix(antigravity): align Hub requests and invalidate versioned model caches

2026-10-20 18:46:02 +08:00

antigravity_executor_disable_cooling_test.go

…

antigravity_executor_execute.go

…

antigravity_executor_finish_reason_test.go

…

antigravity_executor_interactions_test.go

…

antigravity_executor_keepalive_test.go

…

antigravity_executor_request.go

fix(antigravity): align Hub requests and invalidate versioned model caches

2026-10-20 18:46:02 +08:00

antigravity_executor_signature_test.go

…

antigravity_executor_split_usage_test.go

…

antigravity_executor_stream.go

…

antigravity_executor_terminal_disconnect_test.go

…

antigravity_executor_tokens.go

…

antigravity_executor_trailing_reasoning_test.go

…

antigravity_executor_transport_test.go

…

antigravity_executor_unsupported_part_test.go

…

antigravity_executor.go

…

antigravity_home_model_capabilities_test.go

…

antigravity_issue6825_test.go

fix(antigravity): preserve tool call replay continuity and caller isolation

2026-10-21 18:16:58 +08:00

antigravity_preupstream_rewrite_differential_test.go

…

antigravity_preupstream_rewrite_legacy_oracle_test.go

…

antigravity_reasoning_replay_clear_test.go

…

antigravity_reasoning_replay_index_test.go

…

antigravity_reasoning_replay_legacy_oracle_test.go

…

antigravity_reasoning_replay_test.go

…

antigravity_reasoning_replay.go

fix(antigravity): preserve tool call replay continuity and caller isolation

2026-10-21 18:16:58 +08:00

antigravity_refresh_issue6199_test.go

fix(antigravity): align Hub requests and invalidate versioned model caches

2026-10-20 18:46:02 +08:00

antigravity_refresh_test.go

…

antigravity_schema_sanitize_test.go

…

antigravity_transport_decorator_test.go

…

apply_patch_bridge_test.go

…

apply_patch_capability_test.go

…

apply_patch_identity_test.go

…

apply_patch_integration_test.go

…

apply_patch_repair_test.go

…

apply_patch_source_stop_test.go

…

caching_verify_test.go

fix(claude): drop empty text blocks and skip them during cache_control injection

2026-10-21 08:44:10 +08:00

claude_cloaked_cache_repro_test.go

…

claude_cloaked_date_pin_test.go

…

claude_effort_directive_test.go

…

claude_executor_auth_race_test.go

…

claude_executor_auth_test.go

…

claude_executor_auth.go

…

claude_executor_beta_passthrough_test.go

…

claude_executor_beta_policy_test.go

…

claude_executor_cloaking_display_test.go

…

claude_executor_cloaking.go

fix(claude): drop empty text blocks and skip them during cache_control injection

2026-10-21 08:44:10 +08:00

claude_executor_compaction_test.go

…

claude_executor_compaction.go

feat(executor): support Codex remote-compaction v2 on Kimi and Meta Responses routes

2026-10-19 22:07:14 +08:00

claude_executor_diagnostics_test.go

…

claude_executor_diagnostics.go

…

claude_executor_execute.go

fix(claude): map model_context_window_exceeded to length in chat completions

2026-10-22 17:21:21 +08:00

claude_executor_fable_ratelimit_test.go

…

claude_executor_fast_error_test.go

…

claude_executor_fast_error.go

…

claude_executor_native_helper_test.go

…

claude_executor_ratelimit_test.go

…

claude_executor_request_bench_test.go

…

claude_executor_request_remap_test.go

…

claude_executor_request.go

fix(claude): treat sdk-ts as native Claude Code and preserve caller headers under cloak_mode never

2026-10-21 01:11:27 +08:00

claude_executor_stop_reasons_test.go

fix(claude): map model_context_window_exceeded to length in chat completions

2026-10-22 17:21:21 +08:00

claude_executor_stream_terminal_test.go

…

claude_executor_stream_truncation_test.go

…

claude_executor_stream.go

fix(claude): treat sdk-ts as native Claude Code and preserve caller headers under cloak_mode never

2026-10-21 01:11:27 +08:00

claude_executor_subagent_ttl_regression_test.go

…

claude_executor_test.go

fix(claude): treat sdk-ts as native Claude Code and preserve caller headers under cloak_mode never

2026-10-21 01:11:27 +08:00

claude_executor_thinking_signature_test.go

…

claude_executor_tokens.go

fix(claude): treat sdk-ts as native Claude Code and preserve caller headers under cloak_mode never

2026-10-21 01:11:27 +08:00

claude_executor_tool_alias_scope_test.go

…

claude_executor_tool_search_disabled_test.go

…

claude_executor_tool_search_test.go

…

claude_executor_tool_state.go

…

claude_executor_web_search_results_test.go

fix(claude): scope web search result cache by upstream credential

2026-10-21 00:23:55 +08:00

claude_executor_wire_casing_test.go

…

claude_executor.go

fix(claude): scope web search result cache by upstream credential

2026-10-21 00:23:55 +08:00

claude_fingerprint_policy_test.go

feat(claude): strip Claude Code identity on API keys with cloak mode never or uncloak

2026-10-19 12:59:13 +08:00

claude_fingerprint_policy.go

…

claude_header_overrides_issue5093_test.go

…

claude_issue_6120_test.go

…

claude_issue_6193_test.go

…

claude_issue_6698_test.go

…

claude_messages_passthrough_test.go

…

claude_mid_system_model_test.go

…

claude_prompt_cache_options_test.go

…

claude_request_rate_test.go

…

claude_signing_test.go

…

claude_signing.go

…

claude_thinking_replay_test.go

…

claude_thinking_replay.go

…

claude_tls_handshake_retry_test.go

…

codex_executor_auth_test.go

…

codex_executor_auth.go

feat(management): add read-only SIWC credential status endpoint

2026-10-21 08:07:38 +08:00

codex_executor_cache_test.go

…

codex_executor_claude_effort_test.go

fix(thinking): preserve standalone Claude effort on Codex requests

2026-10-21 20:19:05 +08:00

codex_executor_compact_test.go

fix(codex): reconcile replayed web_search_call history with tool declarations

2026-10-19 19:00:41 +08:00

codex_executor_execute_usage_test.go

…

codex_executor_execute.go

fix(codex): reconcile replayed web_search_call history with tool declarations

2026-10-19 19:00:41 +08:00

codex_executor_grokbuild_keepalive_test.go

…

codex_executor_imagegen_test.go

…

codex_executor_input_ids_test.go

…

codex_executor_instructions_test.go

…

codex_executor_parallel_tool_calls_test.go

…

codex_executor_reasoning_replay_cache_test.go

…

codex_executor_reasoning.go

…

codex_executor_request.go

fix(codex): use original oauth credentials for alpha search

2026-10-21 23:04:23 +08:00

codex_executor_retry_test.go

…

codex_executor_routing_hint_test.go

…

codex_executor_search_test.go

fix(codex): use original oauth credentials for alpha search

2026-10-21 23:04:23 +08:00

codex_executor_signature_test.go

…

codex_executor_siwc_configuration_update_test.go

…

codex_executor_spawn_agent_test.go

…

codex_executor_sse_events_test.go

…

codex_executor_stream_alloc_test.go

…

codex_executor_stream_output_test.go

…

codex_executor_stream.go

fix(codex): reconcile replayed web_search_call history with tool declarations

2026-10-19 19:00:41 +08:00

codex_executor_terminal.go

fix(codex): hold response.metadata during stream bootstrap buffering

2026-10-22 11:55:16 +08:00

codex_executor_tokens_test.go

…

codex_executor_tokens.go

…

codex_executor_tool_schema_test.go

…

codex_executor_translate_test.go

…

codex_executor.go

…

codex_native_fidelity_test.go

…

codex_openai_images_extract_test.go

…

codex_openai_images_test.go

…

codex_openai_images.go

…

codex_per_credential_cloaking_issue6034_test.go

…

codex_response_model_test.go

…

codex_stream_bootstrap_buffering_test.go

fix(codex): hold response.metadata during stream bootstrap buffering

2026-10-22 11:55:16 +08:00

codex_websockets_connection.go

…

codex_websockets_duplex_bootstrap_input_test.go

…

codex_websockets_duplex_credential_failure_test.go

…

codex_websockets_duplex_health_test.go

…

codex_websockets_duplex_initial_failure_test.go

…

codex_websockets_duplex_rejection_test.go

…

codex_websockets_duplex_successor_metadata_test.go

…

codex_websockets_duplex_test.go

…

codex_websockets_duplex.go

fix(openai): reauthorize responses websocket turns and support denied status

2026-10-22 03:41:06 +08:00

codex_websockets_errors.go

…

codex_websockets_execute.go

fix(codex): reconcile replayed web_search_call history with tool declarations

2026-10-19 19:00:41 +08:00

codex_websockets_executor_store_test.go

…

codex_websockets_executor_test.go

…

codex_websockets_executor.go

…

codex_websockets_interrupt_test.go

…

codex_websockets_late_interrupt_test.go

…

codex_websockets_request.go

…

codex_websockets_routing_hint_test.go

…

codex_websockets_session.go

…

codex_websockets_spawn_agent_test.go

…

codex_websockets_stream.go

fix(codex): reconcile replayed web_search_call history with tool declarations

2026-10-19 19:00:41 +08:00

custom_magic_headers_test.go

feat(executor): pass through downstream User-Agent and set default xAI User-Agent

2026-10-20 17:05:03 +08:00

devin_executor_test.go

…

devin_executor.go

…

devin_quota_cooldown_regression_test.go

…

devin_stream_content_test.go

…

devin_user_turn_test.go

…

executor_payload_optimization_test.go

…

gemini_executor_signature_test.go

…

gemini_executor_terminal_disconnect_test.go

fix(gemini): preserve usage after terminal stream cancellation

2026-10-21 10:33:05 +08:00

gemini_executor_test.go

…

gemini_executor.go

fix(gemini): preserve usage after terminal stream cancellation

2026-10-21 10:33:05 +08:00

gemini_interactions_continuation_test.go

…

gemini_interactions_tool_names_test.go

…

gemini_interactions_translate_test.go

…

gemini_vertex_executor_interactions_test.go

…

gemini_vertex_executor_openapi_settings_test.go

feat(vertex): support per-model and per-credential openapi routing settings

2026-10-21 02:19:42 +08:00

gemini_vertex_executor_openapi_test.go

feat(vertex): support per-model and per-credential openapi routing settings

2026-10-21 02:19:42 +08:00

gemini_vertex_executor_terminal_disconnect_test.go

…

gemini_vertex_executor_test.go

…

gemini_vertex_executor_trailing_reasoning_test.go

…

gemini_vertex_executor.go

fix(gemini): preserve usage after terminal stream cancellation

2026-10-21 10:33:05 +08:00

home_codex_terminal_test.go

…

issue6258_terminal_test.go

…

issue6381_responses_eof_test.go

…

issue6478_tool_ids_edges_test.go

…

issue6478_tool_ids_test.go

…

kimi_executor_test.go

feat(executor): support Codex remote-compaction v2 on Kimi and Meta Responses routes

2026-10-19 22:07:14 +08:00

kimi_executor.go

feat(executor): support Codex remote-compaction v2 on Kimi and Meta Responses routes

2026-10-19 22:07:14 +08:00

kimi_home_model_capabilities_test.go

…

kimi_thinking_replay_test.go

…

kimi_thinking_replay.go

…

meta_executor_execute.go

feat(executor): support Codex remote-compaction v2 on Kimi and Meta Responses routes

2026-10-19 22:07:14 +08:00

meta_executor_stream.go

fix(meta): measure streaming TTFT on effective token frames

2026-10-20 16:57:14 +08:00

meta_executor_terminal_disconnect_test.go

…

meta_executor_test.go

feat(executor): support Codex remote-compaction v2 on Kimi and Meta Responses routes

2026-10-19 22:07:14 +08:00

meta_executor_ttft_test.go

fix(executor): count completed reasoning items toward TTFT

2026-10-20 17:17:41 +08:00

meta_executor.go

…

oauth_scope_executor.go

…

oauth_scope_test.go

…

openai_compat_executor_compact_test.go

…

openai_compat_executor_images_test.go

…

openai_compat_executor_max_tokens_test.go

…

openai_compat_executor_reasoning_test.go

…

openai_compat_executor_retry_test.go

…

openai_compat_executor_tool_results_test.go

…

openai_compat_executor_ttft_test.go

…

openai_compat_executor_video_test.go

…

openai_compat_executor.go

feat(executor): pass through downstream User-Agent and set default xAI User-Agent

2026-10-20 17:05:03 +08:00

openai_compat_home_options_test.go

…

openai_responses_signature_test.go

…

openai_responses_signature.go

…

payload_barrier_test.go

…

payload_review_test.go

…

request_proxy_priority_test.go

…

response_model_multiprovider_test.go

…

unsupported_part_executor_test.go

…

vertex_proxy_token_test.go

…

websocket_lifecycle_bind_test.go

…

websocket_proxy_reuse_test.go

…

websocket_session_target_test.go

…

websocket_upstream_disconnect_test.go

…

xai_client_version_test.go

…

xai_configuration_update_plugin_test.go

…

xai_configuration_update_test.go

…

xai_executor_auth.go

…

xai_executor_execute.go

…

xai_executor_media.go

…

xai_executor_request.go

feat(executor): pass through downstream User-Agent and set default xAI User-Agent

2026-10-20 17:05:03 +08:00

xai_executor_response.go

…

xai_executor_speech_test.go

…

xai_executor_speech.go

…

xai_executor_stream.go

…

xai_executor_test.go

feat(executor): pass through downstream User-Agent and set default xAI User-Agent

2026-10-20 17:05:03 +08:00

xai_executor_tokens.go

…

xai_executor.go

feat(executor): pass through downstream User-Agent and set default xAI User-Agent

2026-10-20 17:05:03 +08:00

xai_reasoning_replay.go

…

xai_status_err_test.go

…

xai_websockets_executor_test.go

feat(executor): pass through downstream User-Agent and set default xAI User-Agent

2026-10-20 17:05:03 +08:00

xai_websockets_executor.go

feat(executor): pass through downstream User-Agent and set default xAI User-Agent

2026-10-20 17:05:03 +08:00