The agent encyclopedia.
Find a concept. Understand its role. Connect it to the rest of your stack.
38 topics
What is the OpenAI Agents API?
Understand the managed agent runtime, its core resources, and what your application still owns.
Read topic ComparisonsAgents API vs Agents SDK vs Responses API
Compare runtime ownership, state, tool execution, and the integration work each option requires.
Read topic Getting startedAgents API quickstart
Create your first managed session, stream progress, and check the actual turn outcome.
Read topic Core conceptsSessions, turns & durable state
Understand conversation continuity, steering, saved items, and recovery after a disconnected stream.
Read topic InfrastructureHosted vs self-hosted sandboxes
Choose no environment, OpenAI-hosted compute, or your own execution infrastructure.
Read topic Tools & integrationsConnect MCP servers to Agents API
See where MCP connections run, how authentication works, and how to control exposed tools.
Read topic Tools & integrationsFunction calling & application tools
Connect your own APIs with schemas, required actions, durable results, and explicit authorization.
Read topic Core conceptsMulti-agent orchestration
Understand independent contexts, parallel work, coordination, and when a single agent is enough.
Read topic Integration guidesStreaming, webhooks & event handling
Build progress feedback and reconnect safely without confusing transport state with task success.
Read topic Core conceptsFiles, outputs & published artifacts
Move input files into an environment and retrieve results before resource cleanup.
Read topic Tools & integrationsVaults & MCP authentication
Keep reusable MCP credentials separate from agent definitions and attach them to authorized sessions.
Read topic ProductionAgent security & credential boundaries
Map application keys, executor keys, network access, and approval checks to the correct trust boundary.
Read topic ProductionTracing, usage & debugging
Follow a session from live events to saved history, tool calls, subagent activity, and token usage.
Read topic ProductionAgents API pricing & cost planning
Understand model tokens, tool charges, hosted compute, and application infrastructure costs.
Read topic Getting startedOpenAI Agents API with Python
Install the OpenAI Python SDK and stream a managed agent session from your server.
Read topic Integration guidesAgents API with TypeScript & Next.js
Place the OpenAI client on the server and connect durable agent work to a React interface.
Read topic FoundationsWhat is the OpenAI Agents SDK?
Understand the application-side runner and how it differs from a managed Agents API session.
Read topic FoundationsResponses API in the agent stack
Understand the model-and-tools interface beneath application-controlled agent workflows.
Read topic Tools & integrationsAgent Skills & SKILL.md
Package reusable instructions, scripts, and supporting files for an agent’s task.
Read topic Tools & integrationsProgrammatic Tool Calling
Use JavaScript to coordinate tool calls, filter intermediate data, and return compact results.
Read topic Tools & integrationsRAG, file search & vector stores
Connect private knowledge through retrieval, and distinguish a search index from an agent workspace.
Read topic Tools & integrationsWeb search & source citations
Bring current public information into an agent workflow and preserve attribution in the final answer.
Read topic Integration guidesChatKit & Agent Builder
Understand the chat interface layer and the Agent Builder transition for existing workflows.
Read topic Core conceptsAgent configuration & reusable definitions
Separate reusable behavior from session history, environment settings, and application authorization.
Read topic Tools & integrationsPlugins: connect skills and MCP tools
Understand plugin packaging, capability directories, and the difference between instructions and executable tools.
Read topic InfrastructureSandbox lifecycle & executor connections
Coordinate self-hosted compute startup, reconnection, files, and cleanup without confusing a session with a machine.
Read topic Integration guidesWebhook verification, queues & deduplication
Turn signed session notifications into reliable background work and user-visible status updates.
Read topic ProductionAgent retries, timeouts & failure recovery
Distinguish transport errors, failed turns, disconnected compute, and business effects before retrying work.
Read topic ProductionAgent evaluations & regression testing
Measure task outcomes, tool behavior, safety, and cost before changing an agent workflow.
Read topic FoundationsWhich parts of the agent stack do you need?
Separate required application responsibilities, optional integrations, and alternative orchestration runtimes.
Read topic Integration guidesAgents API contract quick reference
Find the session methods, function-result envelope, event meanings, and retry boundaries for an Agents API integration.
Read topic Integration guidesOrder agent reference: approval, recovery & evaluation
Run a Python and SQLite reference workflow with human approval, a durable tool-result ledger, restart recovery, and nine acceptance tests.
Read topic Integration guidesBuild a Next.js order agent: server, tools & streaming
Run a complete local order-lookup example with server-side credentials, a validated function tool, and browser progress.
Read topic Getting startedPython Agents API example with a real tool loop
Create a session, handle a read-only function call, stream progress, and check the root turn outcome in one Python file.
Read topic ProductionHuman approval & prompt injection defenses
Put authorization and confirmation between an agent's proposal and a real-world action.
Read topic ProductionAgents API troubleshooting: symptoms, checks & recovery
Diagnose access failures, pending tools, missing output, disconnected environments, and duplicate work.
Read topic ComparisonsAgent tool support & runtime ownership matrix
Compare where orchestration, function handlers, MCP connections, and compute belong before reusing an integration.
Read topic Core conceptsAgent resources, relationships & task state
Map configuration, sessions, turns, events, items, environments, and artifacts to their owners and lifetimes.
Read topic