Issues

• Responses incomplete at max_output_tokens: billed usage is dropped from RunContextWrapper.usage and the trace (real api.openai.com case for #4901)

  openai/openai-agents-python#6025 · Filter by author alperbey opened on Nov 15, 2026
  1

• OpenAIResponsesCompactionSession over a limited store sends a function_call_output without its function_call

  openai/openai-agents-python#6007 · Filter by author devteamaegis opened on Nov 14, 2026
  1

• native SDK: a restored Session hides a pending needs_approval call, so the turn waits forever

  openai/openai-agents-python#5995 · Filter by author Nixxx19 opened on Nov 14, 2026

• run_demo_loop never asks for tool approval, the paused call goes out without its output

  openai/openai-agents-python#5992 · Filter by author Nixxx19 opened on Nov 14, 2026
  1

• Expose a public getter for the global tracing-disabled state (e.g. agents.tracing.is_tracing_disabled())

  openai/openai-agents-python#5965 · Filter by author tommasobolis opened on Nov 13, 2026

• ImportError: cannot import name 'Docstring' from 'griffe' (unknown location)

  openai/openai-agents-python#5951 · Filter by author db-hui opened on Nov 13, 2026
  2

• Agent end hooks lose the structured purchase details of an agent-as-tool call

  openai/openai-agents-python#5935 · Filter by author master1018 opened on Nov 13, 2026
  4

• LitellmModel with Anthropic models sends tools without strict (litellm drops it)

  openai/openai-agents-python#5825 · Filter by author Android-Login-Analysis opened on Nov 11, 2026
  5

• [Bug]: Silent hang on Windows during async execution in restricted/sandbox environments (ProactorEventLoop loopback socket block)

  openai/openai-agents-python#5811 · Filter by author Anil-Pradhan-web opened on Nov 11, 2026
  1

• Already-completed sibling output guardrail result omitted from tripwire run_data.output_guardrail_results when the tripping result is consumed first (output counterpart of #5567)

  openai/openai-agents-python#5743 · Filter by author reeve25 opened on Nov 10, 2026
  1

• After a RunState JSON round trip, the next nested handoff embeds the previous history summary instead of flattening it

  openai/openai-agents-python#5678 · Filter by author adityasingh2400 opened on Nov 10, 2026
  2

• OpenAIConversationsSession.get_items() and pop_item() return async_ instead of the async wire field on function calls, so replayed or restored history sends an unknown parameter

  openai/openai-agents-python#5670 · Filter by author reeve25 opened on Nov 9, 2026
  3

• Optional A2A support alongside agents.mcp

  enhancement
  openai/openai-agents-python#5660 · Filter by author guglielmo-san opened on Nov 9, 2026
  1

• WebSocket 17 support

  enhancement
  openai/openai-agents-python#5653 · Filter by author attilaolah opened on Nov 9, 2026
  2

• conversation_id run stopping at max_turns leaves tool calls without outputs; every later turn 400s

  feature:core
  openai/openai-agents-python#5642 · Filter by author jaeminbird opened on Nov 9, 2026
  1

• SessionSettings(limit=N) window can start inside a reasoning response, so the Responses API rejects the turn

  feature:sessions
  openai/openai-agents-python#5507 · Filter by author jaeminbird opened on Nov 4, 2026
  7

• List Sprites as a hosted sandbox platform

  documentationfeature:sandboxes
  openai/openai-agents-python#5498 · Filter by author mattp-fly opened on Nov 4, 2026
  1

• Voice: default sentence splitter never splits Chinese/Japanese, so TTS waits for the whole reply

  feature:voice
  openai/openai-agents-python#5475 · Filter by author jaeminbird opened on Nov 2, 2026
  2

• AnyLLMModel chat completions: extra_headers TypeError on Mistral (Chat.complete_async)

  feature:extensions
  openai/openai-agents-python#5473 · Filter by author TehilaTheStudent opened on Nov 2, 2026
  2

• OpenAIChatCompletionsModel sends max_tokens, which reasoning models reject

  openai/openai-agents-python#5455 · Filter by author swf1996120 opened on Oct 30, 2026
  4

• Chat Completions models drop the conversation history when conversation_id/previous_response_id are set

  openai/openai-agents-python#5454 · Filter by author BlueX888 opened on Oct 30, 2026
  2

• Feature proposal: inherited recursion-depth budget for Agent.as_tool delegation

  openai/openai-agents-python#5453 · Filter by author BlackRonin-coder opened on Oct 29, 2026
  9

• Streamed Chat Completions turns report response_id '__fake_id__' instead of None, and it leaks into previous_response_id and compaction

  openai/openai-agents-python#5451 · Filter by author v0ropaev opened on Oct 29, 2026
  3

• Pattern: deterministic self-verification of numeric claims via a function tool

  openai/openai-agents-python#5450 · Filter by author PHM-Audit opened on Oct 29, 2026

• function_tool rejects valid arguments when a parameter uses Pydantic strict mode (Enum, date, UUID)

  openai/openai-agents-python#5448 · Filter by author TheSaiEaranti opened on Oct 28, 2026
  2