OpenAI conversation state guide
Documents multi-turn conversation state, durable conversation identifiers, messages, tool calls, tool outputs, and response chaining across turns.
What this source can teach
Use this source to understand a platform convention and the setting it was designed for.
- Start with the original. Open the source and find the guidance or example relevant to your task.
- Trace the connection. Use the claim notes below to see how this site applies the source to specific patterns.
- Check the setting. Compare the source's platform, people, and constraints with your own decision.
Pattern Decisions This Source Supports
| Pattern | Supported decision | Required contract | Claim note |
|---|---|---|---|
| Agent plan preview | Choose agent plan preview when an AI agent's proposed multi-step execution needs review before it starts. | The preview names the objective, plan version, source context, model or workflow instructions, planned tools, and expected output before run. | Supports state, messages, tool calls, tool outputs, and response chaining that affect agent plans. |
| AI agent acts without approval | Flag this anti-pattern when an AI agent or automation executes a high-impact side effect without showing and requiring approval for the exact action and payload first. | The agent distinguishes read-only steps, draft steps, reversible local changes, and external side-effect steps before execution. | Supports durable conversation, tool calls, tool outputs, and response chaining that affect agent state and side effects. |
| Chat interface | Choose chat interface when the user task depends on turn-taking, follow-up prompts, assistant responses, and conversation history rather than a single request field. | Every user-visible answer is tied to the exact user message and visible context that produced it. | Supports multi-turn conversation state, durable conversation identifiers, messages, tool calls, tool outputs, and response chaining. |
| Correction feedback | Choose correction feedback when users need to correct AI output, source use, assumptions, recommendations, safety behavior, or answer quality after an AI response is shown. | A correction feedback action preserves answer ID, response version, claim span, source ID, user reason, expected correction, submitter, timestamp, and chosen scope. | Supports preserving conversation, message, response, and tool identity for attaching correction feedback. |
| Regenerate / retry | Choose regenerate / retry when the task is rerunning an AI response attempt, creating a new answer version, continuing a failed generation, or recovering from failed AI source or tool work. | Retry same prompt repeats the submitted prompt and visible context unless the UI explicitly shows which context changed. | Supports preserving conversation, prompt, response, and tool state across response attempts. |
| Scope clarification | Choose scope clarification when the AI needs a missing boundary before producing a reliable answer, plan, retrieval query, or tool action. | The system identifies what boundary is missing before asking the user to clarify. | Supports preserving original request and clarification answer across AI turns. |
| Tool-use visibility | Choose tool-use visibility when users need to inspect exact tool names, purposes, inputs, outputs, permissions, side effects, failures, or redactions. | Each tool-use item identifies one tool call or tool-call attempt with stable call ID, run ID, step ID, tool name, status, timestamp, and permission scope. | Supports exposing tool calls, tool outputs, and response chaining as structured conversational state. |
Comparisons supported by this source
- Agent plan preview vs task list vs recommended next action vs human approval gate vs step navigation
Supports multi-turn context, messages, tool calls, tool outputs, and response chaining that affect agent planning state.
- AI agent acts without approval vs human approval gate vs agent plan preview vs agent progress trace vs tool-use visibility vs dangerous-action review
Supports durable agent state, tool calls, tool outputs, and response chaining that must remain tied to authorization context.
- Chat interface vs prompt box vs threaded discussion vs comments vs feed
Supports multi-turn conversation state, conversation identifiers, messages, tool calls, tool outputs, and response chaining.
- Correction feedback vs query correction vs inline validation vs comments vs regenerate / retry vs editable AI output
Supports preserving response and conversation identity so corrections attach to the right AI output.
- Regenerate / retry vs retry vs chat interface vs streaming response vs prompt box vs editable AI output
Supports preserving conversation, prompt, response, and tool state across attempts.
- Scope clarification vs prompt box vs search scope selector vs agent plan preview vs human approval gate vs chat interface
Supports preserving request and clarification state across AI turns.
- Tool-use visibility vs agent progress trace vs source grounding display vs citation display vs activity log vs human approval gate
Supports structured tool calls and tool outputs in conversation state.
Evidence Role
This source is treated as platform-guideline evidence. Its value comes from the specific claims and setting described above.
Publisher: OpenAI. Last checked: .