Agents：OpenAI Agents SDK

* [Dynamic instructions](https://openai.github.io/openai-agents-python/agents/#dynamic-instructions). * [Forcing tool use](https://openai.github.io/openai-agents-python/agents/#forcing-tool-use). …

* [Dynamic instructions](https://openai.github.io/openai-agents-python/agents/#dynamic-instructions). * [Forcing tool use](https://openai.github.io/openai-agents-python/agents/#forcing-tool-use). * [Tool use behavior](https://openai.github.io/openai-agents-python/agents/#tool-use-behavior). * [Tools](https://openai.github.io/openai-agents-python/tools/). * [Agents module](https://openai.github.io/openai-agents-python/ref/). * [Agent](https://openai.github.io/openai-agents-python/ref/agent/). * [Runner](https://openai.github.io/openai-agents-python/ref/run/). * [Run config](https://openai.github.io/openai-agents-python/ref/run_config/). * [Run state](https://openai.github.io/openai-agents-python/ref/run_state/). * [Overview](https://openai.github.io/openai-agents-python/ref/sandbox/). * [Manifest](https://openai.github.io/openai-agents-python/ref/sandbox/manifest/). * [Permissions](https://openai.github.io/openai-agents-python/ref/sandbox/permissions/). * [SnapshotSpec](https://openai.github.io/openai-agents-python/ref/sandbox/snapshot/). * [Memory](https://openai.github.io/openai-agents-python/ref/sandbox/capabilities/memory/). * [Memory](https://openai.github.io/openai-agents-python/ref/memory/). * [REPL](https://openai.github.io/openai-agents-python/ref/repl/). * [Tools](https://openai.github.io/openai-agents-python/ref/tool/). * [Tool context](https://openai.github.io/openai-agents-python/ref/tool_context/). * [Results](https://openai.github.io/openai-agents-python/ref/result/). * [Handoffs](https://openai.github.io/openai-agents-python/ref/handoffs/). * [Lifecycle](https://openai.github.io/openai-agents-python/ref/lifecycle/). * [Items](https://openai.github.io/openai-agents-python/ref/items/). * [Run context](https://openai.github.io/openai-agents-python/ref/run_context/). * [Usage](https://openai.github.io/openai-agents-python/ref/usage/). * [Exceptions](https://openai.github.io/openai-agents-python/ref/exceptions/). * [Guardrails](https://openai.github.io/openai-agents-python/ref/guardrail/). * [Prompts](https://openai.github.io/openai-agents-python/ref/prompts/). * [Model settings](https://openai.github.io/openai-agents-python/ref/model_settings/). * [Tool guardrails](https://openai.github.io/openai-agents-python/ref/tool_guardrails/). * [Computer](https://openai.github.io/openai-agents-python/ref/computer/). * [Agent output](https://openai.github.io/openai-agents-python/ref/agent_output/). * [Function schema](https://openai.github.io/openai-agents-python/ref/function_schema/). * [Scope](https://openai.github.io/openai-agents-python/ref/tracing/scope/). * [RealtimeRunner](https://openai.github.io/openai-agents-python/ref/realtime/runner/). * [RealtimeSession](https://openai.github.io/openai-agents-python/ref/realtime/session/). * [Events](https://openai.github.io/openai-agents-python/ref/realtime/events/). * [Configuration](https://openai.github.io/openai-agents-python/ref/realtime/config/). * [Model](https://openai.github.io/openai-agents-python/ref/realtime/model/). * [Pipeline](https://openai.github.io/openai-agents-python/ref/voice/pipeline/). * [Workflow](https://openai.github.io/openai-agents-python/ref/voice/workflow/). * [Input](https://openai.github.io/openai-agents-python/ref/voice/input/). * [Result](https://openai.github.io/openai-agents-python/ref/voice/result/). * [Events](https://openai.github.io/openai-agents-python/ref/voice/events/). * [Exceptions](https://openai.github.io/openai-agents-python/ref/voice/exceptions/). * [Model](https://openai.github.io/openai-agents-python/ref/voice/model/). * [Utils](https://openai.github.io/openai-agents-python/ref/voice/utils/). * [Handoff prompt](https://openai.github.io/openai-agents-python/ref/extensions/handoff_prompt/). * [Any-LLM model](https://openai.github.io/openai-agents-python/ref/extensions/models/any_llm_model/). * [Any-LLM provider](https://openai.github.io/openai-agents-python/ref/extensions/models/any_llm_provider/). * [Tool output trimmer](https://openai.github.io/openai-agents-python/ref/extensions/tool_output_trimmer/). * [Dynamic instructions](https://openai.github.io/openai-agents-python/agents/#dynamic-instructions). * [Forcing tool use](https://openai.github.io/openai-agents-python/agents/#forcing-tool-use). * [Tool use behavior](https://openai.github.io/openai-agents-python/agents/#tool-use-behavior). See [Dynamic instructions](https://openai.github.io/openai-agents-python/agents/#dynamic-instructions). See [Tools](https://openai.github.io/openai-agents-python/tools/). See [Tool use behavior](https://openai.github.io/openai-agents-python/agents/#tool-use-behavior). See [Forcing tool use](https://openai.github.io/openai-agents-python/agents/#forcing-tool-use). * LLM, tool, and handoff hooks receive [`RunContextWrapper`](https://openai.github.io/openai-agents-python/ref/run_context/#agents.run_context.RunContextWrapper “RunContextWrapper. You can force tool use by setting [`ModelSettings.tool_choice`](https://openai.github.io/openai-agents-python/ref/model_settings/#agents.model_settings.ModelSettings.tool_choice “tool_choice.

AI Agent行业深度分析

1. Agents SDK | OpenAI API

* [Using tools](https://developers.openai.com/api/docs/guides/tools). * [Overview](https://developers.openai.com/api/docs/guides/agents). * [Quickstart](https://developers.openai.com/api/docs/guides/agents/quickstart). * [Models and providers](https://developers.openai.com/api/docs/guides/agents/models). * [Orchestration](https://developers.openai.com/api/docs/guides/agents/orchestration). * [Results and state](https://developers.openai.com/api/docs/guides/agents/results). * [Evaluate agent workflows](https://developers.openai.com/api/docs/guides/agent-evals). * [Overview](https://developers.openai.com/api/docs/guides/agent-builder). * [Overview](https://developers.openai.com/api/docs/guides/chatkit). * [Skills](https://developers.openai.com/api/docs/guides/tools-skills). * [Shell](https://developers.openai.com/api/docs/guides/tools-shell). * [Computer use](https://developers.openai.com/api/docs/guides/tools-computer-use). * [Retrieval](https://developers.openai.com/api/docs/guides/retrieval). * [Webhooks](https://developers.openai.com/api/docs/guides/webhooks). * [Compaction](https://developers.openai.com/api/docs/guides/compaction). * [Overview](https://developers.openai.com/api/docs/guides/prompting). * [Reasoning models](https://developers.openai.com/api/docs/guides/reasoning). * [Getting started](https://developers.openai.com/api/docs/guides/evaluation-getting-started). * [Overview](https://developers.openai.com/api/docs/guides/realtime). * [Graders](https://developers.openai.com/api/docs/guides/graders). * [Embeddings](https://developers.openai.com/api/docs/guides/embeddings). * [Moderation](https://developers.openai.com/api/docs/guides/moderation). * [Overview](https://developers.openai.com/api/docs/guides/latency-opt

查看完整报道 →

2. A practical guide to building agents | OpenAI

[Skip to main content](https://openai.com/business/guides-and-resources/a-practical-guide-to-building-ai-agents/#main). * [Introduction](https://openai.com/business/guides-and-resources/a-practical-guide-to-building-ai-agents/#introduction). * [What is an agent?](https://openai.com/business/guides-and-resources/a-practical-guide-to-building-ai-agents/#what-is-an-agent). * [When should you build an agent?](https://openai.com/business/guides-and-resources/a-practical-guide-to-building-ai-agents/#when-should-you-build-an-agent). * [Agent design foundations](https://openai.com/business/guides-and-resources/a-practical-guide-to-building-ai-agents/#agent-design-foundations). * [Guardrails](https://openai.com/business/guides-and-resources/a-practical-guide-to-building-ai-agents/#guardrails). * [Conclusion](https://openai.com/business/guides-and-resources/a-practical-guide-to-building-ai-agents/#conclusion). 1. **Single-agent systems,**where a single model equipped with appropriate tools and instructions executes workflows in a loop. `1from agents import Agent, Runner23technical_support_agent = Agent(4 name=”Technical Support Agent”,5 instructions=(6 “You provide expert assistance with resolving technical issues, “7 “system outages, or product troubleshooting.”8 ),9 tools=[search_knowledge_base],10)1112sales_assistant_agent = Agent(13 name=”Sales Assistant Agent”,14 instructions=(15 “You help enterprise clients browse the product catalog, “16 “recommend suitable solutions, and facilitate purchase transactions.”17 ),18 tools=[initiate_purchase_order],19)2021order_management_agent = Agent(22 name=”Order Management Agent”,23 instructions=(24 “You assist clients with inquiries regarding order tracking, “

查看完整报道 →

3. Solutions for agentic workflows

Try now(opens in a new window)Contact sales. Deploy agents that can reason, take action, and assist across a wide range of work—from simple tasks to complex projects. Image 1: Workflow interface showing a connected sequence starting with a green “Start” button leading to a blue “Categorize Agent” step on a soft blue and orange gradient background. Explore OpenAI products that help you use, build, and scale agentic AI. Image 2: ChatGPT interface with browser tabs showing Aerion shoe pages reopened after a user request, with a note confirming four tabs opened and an option to pull more from the last month. * Agents follow your team’s rules and act with the right approvals. Try ChatGPT Business(opens in a new window). Start building(opens in a new window). * Explore ChatGPT(opens in a new window). * API log in(opens in a new window). (opens in a new window)(opens in a new window)(opens in a new window)(opens in a new window)(opens in a new window)(opens in a new window)(opens in a new window).

查看完整报道 →

4. The next evolution of the Agents SDK – OpenAI

[Skip to main content](https://openai.com/index/the-next-evolution-of-the-agents-sdk/#main). * [Developers](https://openai.com/api/). * [A more capable harness for the agent loop](https://openai.com/index/the-next-evolution-of-the-agents-sdk/#a-more-capable-harness-for-the-agent-loop). * [Native sandbox execution](https://openai.com/index/the-next-evolution-of-the-agents-sdk/#native-sandbox-execution). * [Separating harness from compute for security, durability, and scale](https://openai.com/index/the-next-evolution-of-the-agents-sdk/#separating-harness-from-compute-for-security-durability-and-scale). * [Pricing and availability](https://openai.com/index/the-next-evolution-of-the-agents-sdk/#pricing-and-availability). * [What’s next](https://openai.com/index/the-next-evolution-of-the-agents-sdk/#whats-next). We’re introducing new capabilities to the [Agents SDK⁠(opens in a new window)](https://developers.openai.com/api/docs/guides/agents) that give developers standardized infrastructure that is easy to get started with and is built correctly for OpenAI models: a model-native harness that lets agents work across files and tools on a computer, plus native sandbox execution for running that work safely. These primitives include tool use via [MCP⁠(opens in a new window)](https://modelcontextprotocol.io/), progressive disclosure via [skills⁠(opens in a new window)](https://agentskills.io/), custom instructions via [AGENTS.md⁠(opens in a new window)](https://agents.md/), code execution using the [shell⁠(opens in a new window)](https://developers.openai.com/api/docs/guides/tools-shell) tool, file edits using the [apply patch⁠(opens in a new window)](https://developers.openai.com/api/docs/guides/tools-apply-patch) tool, and more. * [Documentation(opens in a new wi

查看完整报道 →