openai-agents-python

Learn how to implement multi-agent workflows with agent-to-agent delegation in openai-agents-python. Master handoff and context transfer for seamless agent collaboration.

Easily migrate from Chat Completions to the Responses API in openai-agents-python. Update your default API setting once and simplify your provider instances for a smoother workflow.

Learn to use CodeInterpreterTool for secure, sandboxed code execution within openai-agents-python. Empower your agents with computational capabilities while protecting your host system.

Learn to enable and configure tracing sensitive data exclusion in openai-agents-python. Secure your traces by setting environment variables or run config options.

Learn to use WebSearchTool and FileSearchTool in openai-agents-python for effective information retrieval. Integrate these tools into your agent setup for powerful querying.

Manage conversation history in openai-agents-python using handoff history mappers. Collapse transcripts into one message or customize context for the next agent.

Master browser automation with ComputerTool and AsyncComputer in openai-agents-python. Control real browsers for navigation, clicking, typing, and screenshots. Learn how today.

Master error handling in openai-agents-python with RunErrorHandlers. Intercept errors like MaxTurnsExceeded and customize outputs for a smoother user experience.

Discover the difference between tool input/output guardrails and regular guardrails in openai-agents-python. Gain granular control over agent behavior and content rejection without halting the process.

Intercept agent lifecycle events in openai-agents-python using RunHooks and AgentHooks. Easily customize agent behavior without altering core logic. Learn how now!

Learn to stream agent responses and handle streaming events with openai-agents-python. Use Runner.run_streamed() to access typed events like LLM deltas, tool calls, and agent handoffs.

Learn to configure ModelRetrySettings and custom retry policies in openai-agents-python. Control failed model calls with max retries, backoff, and custom policies.