9 Pull requests merged by 8 people

• Fix #976 MCP filtering, make agent and run ctx optional

  #977 merged Jul 8, 2025

• avoid concat tool call delta id

  #990 merged Jul 8, 2025

• description_override is not properly used for function_schema.description

  #1000 merged Jul 8, 2025

• Support for file_input content (#fix 893)

  #1009 merged Jul 8, 2025

• Add Makefile check rule to run all the linters and checkers

  #1038 merged Jul 8, 2025

• Add on_start support to VoiceWorkflowBase and VoicePipeline

  #922 merged Jul 8, 2025

• Support mcp prompts

  #1010 merged Jul 8, 2025

• Fix agent lifecycle example output to reflect correct hook execution order

  #1015 merged Jul 8, 2025

• docs: mention 'name' is required in Agent constructor . fix: mention 'name' as required in Agent init doc

  #1033 merged Jul 8, 2025

8 Pull requests opened by 7 people

• Support OPENAI_BASE_URL in tracing

  #997 opened Jul 2, 2025

• Added runtime validation for Agent name field

  #998 opened Jul 3, 2025

• Support Streaming of Function Call Arguments

  #999 opened Jul 3, 2025

• fix(hooks): handle kwargs in hook calls

  #1002 opened Jul 3, 2025

• Fix: Ensure output_guardrail provides function name instead of defaulting to None

  #1003 opened Jul 3, 2025

• Add run configuration inheritance for agent-as-tool workflows

  #1006 opened Jul 4, 2025

• Fix #968 by upgrading openai package to the latest

  #1034 opened Jul 8, 2025

• feature/session memory

  #1040 opened Jul 8, 2025

29 Issues closed by 12 people

• Add list_prompts / get_prompt to MCP server implementations

  #1011 closed Jul 8, 2025

• [BUG]: MCP Server unecessarily requires use of Agent and RunContext

  #976 closed Jul 8, 2025

• OpenAI 'file' message type is supported by Chat Completions

  #893 closed Jul 8, 2025

• Add Intro message function for VoicePipeline

  #488 closed Jul 8, 2025

• 500 InternalServerError from Google Generative AI client with no retry/backoff

  #1014 closed Jul 8, 2025

• `str` allows int value without error in Agent constructor

  #1032 closed Jul 8, 2025

• How system prompt constructed?

  #1028 closed Jul 8, 2025

• Unable to override 5 s MCP tool timeout with documented env-vars on v0.1.0

  #1026 closed Jul 8, 2025

• Error while trying to use Custom function tools

  #1023 closed Jul 8, 2025

• Typo in Agent constructor docs—name marked required but None, int allowed

  #1019 closed Jul 8, 2025

• agents.exceptions.UserError: Hosted tools are not supported with the ChatCompletions API.

  #1008 closed Jul 8, 2025

• Failed to run when trying the example financial_research_agent with qwen3

  #975 closed Jul 8, 2025

• MCPServerSse not able to connect to mcp server running within docker container

  #946 closed Jul 8, 2025

• Clarify how to filter final vs. intermediate generation events in streaming API

  #947 closed Jul 8, 2025

• Potentially Adding Citation feature button for Researchers

  #949 closed Jul 8, 2025

• Typed handoffs

  #955 closed Jul 8, 2025

• Support for passing extra headers in request for LitellmModel Class

  #1031 closed Jul 7, 2025

• How to deal with agents not calling the appropriate tools

  #178 closed Jul 7, 2025

• Feature Request: Add a configurable timeout for MCP tool calls

  #1024 closed Jul 7, 2025

• Support Anthropic Prompt Caching

  #905 closed Jul 7, 2025

• Adding a ToolRAG that the user can enable when he gives to the agent many tools.

  #913 closed Jul 7, 2025

• Adding a knowledge parameter within the Agent class

  #914 closed Jul 7, 2025

• Go version of the Agents SDK?

  #942 closed Jul 7, 2025

• Provided By OpenAI Kittens SSO

  #929 closed Jul 6, 2025

• Document upload across gemini, claude and gpt (completions API)

  #933 closed Jul 6, 2025

• [Docs Bug] RunConfig.model does not override Agent.model as stated in documentation

  #1007 closed Jul 4, 2025

• chat completion model failed to produce multiple tool call items

  #989 closed Jul 2, 2025

• Open ai agent python

  #983 closed Jul 2, 2025

• docs/mcp.md

  #988 closed Jul 1, 2025

20 Issues opened by 20 people

• Guidance on using the same agent instance across multiple requests

  #1039 opened Jul 8, 2025

• Tool Bug

  #1037 opened Jul 8, 2025

• Reasoning only coming before final output

  #1036 opened Jul 8, 2025

• Empty tool output incorrectly triggers “Errored MCP tool result”

  #1035 opened Jul 8, 2025

• Add `tool_name` to `ToolContext` for generic tool handlers

  #1030 opened Jul 7, 2025

• MCP Streamable HTTP Server Client times out

  #1029 opened Jul 7, 2025

• reasoning_item_created never called on stream

  #1025 opened Jul 7, 2025

• Passing Local Context to MCP Server Tools

  #1021 opened Jul 6, 2025

• Incomplete RunItemStreamEvent names list

  #1020 opened Jul 6, 2025

• Feature Request: Add "include_domains" and "exclude_domains" parameters to "WebSearchTool"

  #1018 opened Jul 6, 2025

• Add support for reference images in Agents SDK for image generation

  #1017 opened Jul 6, 2025

• Unclosed client session when using some litellm models

  #1016 opened Jul 5, 2025

• Tool Call Execution Problem with Structured Output

  #1012 opened Jul 5, 2025

• agents.exceptions.UserError: Unhandled item type or structure when use reasoning model with tool call.

  #1004 opened Jul 4, 2025

• How should I handle when the response from MCP or some tool is JSON, which contains the base64 encoding of the file.

  #1001 opened Jul 3, 2025

• No validation on the Agent constructor arguments

  #996 opened Jul 2, 2025

• Converter.items_to_messages cannot handle assistant message with both content and tool_calls

  #995 opened Jul 2, 2025

• Agent unable to perform multiple Web Searches by using WebSearchTool

  #993 opened Jul 2, 2025

• FunctionTool ignores strict_json_schema parameter

  #992 opened Jul 2, 2025

• Input guardrail tripwire allows tool execution to continue after exception is raised

  #991 opened Jul 2, 2025

38 Unresolved conversations

Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.

• Add logprobs to ModelSettings

  #971 commented on Jul 8, 2025 • 3 new comments

• Add Sessions for Automatic Conversation History Management

  #752 commented on Jul 8, 2025 • 1 new comment

• Cannot run visualization

  #978 commented on Jul 8, 2025 • 0 new comments

• Function tool choice causing pydantic validation error

  #980 commented on Jul 8, 2025 • 0 new comments

• Agent keep retrying until hit max_retires on function calling

  #981 commented on Jul 8, 2025 • 0 new comments

• Agent as a tool input_type

  #984 commented on Jul 8, 2025 • 0 new comments

• Error Using Gemini API Key with Agent-as-Tool Pattern in openai-agent-sdk

  #986 commented on Jul 8, 2025 • 0 new comments

• Support Streaming of Function Call Arguments

  #834 commented on Jul 8, 2025 • 0 new comments

• Error "Item ‘rs_ABCD’ of type ‘reasoning’ was provided without its required..." when using CodeInterpreter

  #985 commented on Jul 8, 2025 • 0 new comments

• Added support for gpt4o-realtime models for Speect to Speech interactions

  #659 commented on Jul 5, 2025 • 0 new comments

• feat(tools): run sync tools in a thread to avoid event loop blocking

  #820 commented on Jul 4, 2025 • 0 new comments

• Added support for "return" handoffs (#1)

  #869 commented on Jul 7, 2025 • 0 new comments

• Ensure that input_guardrails can block tools from running

  #931 commented on Jul 6, 2025 • 0 new comments

• Support Claude extended thinking

  #941 commented on Jul 6, 2025 • 0 new comments

• Fix critical tracing module issues and improve reliability

  #962 commented on Jul 8, 2025 • 0 new comments

• chore(vscode): upgrade .vscode/settings.json to full‑stack DX config

  #964 commented on Jul 8, 2025 • 0 new comments

• Feat: separate tool_call_item and tool_call_output_item in stream events

  #974 commented on Jul 8, 2025 • 0 new comments

• Add MCP Tools to visualization.

  #979 commented on Jul 8, 2025 • 0 new comments

• feat(agents): Add on_llm_start and on_llm_end Lifecycle Hooks

  #987 commented on Jul 8, 2025 • 0 new comments

• Custom model provider ignored when using agents as tools

  #663 commented on Jul 2, 2025 • 0 new comments

• Error initializing MCP server: Connection closed

  #822 commented on Jul 3, 2025 • 0 new comments

• KeyError: ~TContext while importing with Python 3.11.0

  #891 commented on Jul 3, 2025 • 0 new comments

• Can tools ingest images attached to image_url?

  #875 commented on Jul 3, 2025 • 0 new comments

• Invalid parameter: messages with role 'tool' must be a response to a preceeding message with 'tool_calls'.

  #873 commented on Jul 3, 2025 • 0 new comments

• OPENAI Authentication error

  #870 commented on Jul 3, 2025 • 0 new comments

• A2A (Agent2Agent) support

  #472 commented on Jul 3, 2025 • 0 new comments

• Support for short‑term and long‑term memory in Agents SDK

  #887 commented on Jul 4, 2025 • 0 new comments

• Connection Error on Windows When Running Official LiteLLM Example

  #836 commented on Jul 4, 2025 • 0 new comments

• Human-In-The-Loop Architecture should be implemented on top priority!

  #636 commented on Jul 4, 2025 • 0 new comments

• Tool calling with LiteLLM and thinking models fail

  #765 commented on Jul 4, 2025 • 0 new comments

• Support for MCP prompts and resources

  #544 commented on Jul 5, 2025 • 0 new comments

• results field in Code Interpreter always returns None

  #917 commented on Jul 5, 2025 • 0 new comments

• When will realtime capabilities be added like they are in the Typescript SDK?

  #894 commented on Jul 5, 2025 • 0 new comments

• OpenAI SDK does not expose MCP session ID, making stateful streamable HTTP sessions impossible

  #924 commented on Jul 5, 2025 • 0 new comments

• separate stream items for "tool_call_item" and "tool_call_output_item"

  #831 commented on Jul 8, 2025 • 0 new comments

• Pydantic requirements mismatch

  #967 commented on Jul 8, 2025 • 0 new comments

• pydantic_core._pydantic_core.ValidationError: 1 validation error for ResponseFunctionWebSearch action

  #968 commented on Jul 8, 2025 • 0 new comments

• Handoff to agent that uses websearch tool call fails. But works fine if agent is called as a tool

  #972 commented on Jul 8, 2025 • 0 new comments