The OpenAI agent ecosystem
Understand the stack.
Build better agents.
Explore how the Agents API, models, tools, and infrastructure connect. From your first session to production.
Explore the agent ecosystem
Ecosystem mapClick a component to explore
tasks ↓ ↑ events
Sessions · Context · Orchestration
inference
tool calls
executes in
results & saved work ↓
Solid connections show dependenciesEnvironment is optional
Component overview
Agents API
OpenAI runs the Codex harness: the model-and-tool loop, durable sessions, context compaction, and recovery.
- UPSTREAM
- Application requests and configuration
- DOWNSTREAM
- Models, tools, optional environments
Based on OpenAI documentation View sources
A clear path through the ecosystem
Start with the big picture, choose your runtime, or go straight to code.
Get the big picture
Agents, sessions, tools, and environments. Understand what each piece does.
Explore the architectureChoose your runtime
Agents API, Agents SDK, or Responses? Compare who runs what.
Compare the optionsMake your first call
Install the SDK, create a session, and follow the agent’s progress.
Open the quickstartWhat happens after you send a task?
Walk through the request, tool call, and result one event at a time.
ILLUSTRATIVE SESSION FLOW
Submit a task
Your server creates a session and supplies instructions, input, and permitted tools.
Application → Agents APIGo deeper into the stack
Integration guides
Build a Next.js order agent: server, tools & streaming
Core concepts
Agent resources, relationships & task state
Comparisons
Agent tool support & runtime ownership matrix
Production
Agents API troubleshooting: symptoms, checks & recovery
Tools & integrations
Connect MCP servers to Agents API
Infrastructure
Hosted vs self-hosted sandboxes
Core concepts
Multi-agent orchestration
Production
Agent security & credential boundaries
Core concepts
Files, outputs & published artifacts
Production