Build Durable AI Agents that survive failures mid-conversation, or hand your AI coding assistant deep Temporal expertise.
Long-running agents need durable state, tool-call retries, and the ability to resume mid-conversation when a Worker crashes. Follow these tutorials to learn how to add Temporal to your applications.
Build a durable AI application backed by Temporal's reliable execution model.
Wrap Temporal Workflows as MCP tools so LLM agents can call them durably.
Use Temporal and the OpenAI Agents SDK to build deep-research agents that run for hours and survive failure.
Compose Temporal primitives into a long-running agent that picks up where it left off.
Bite-sized Python recipes you can copy into a project today. Each one shows a single pattern - calling an LLM, looping with tool calls, signaling for human approval - and explains the Temporal design decisions behind it.
Call an LLM from a Workflow. The Temporal-flavored Hello World - generic Activity, retries handled by Temporal, not the OpenAI client.
The foundational tool-calling pattern: LLM decides what tool to invoke, Temporal Activity runs it, result feeds back to the LLM.
Full agentic loop with dynamic Activities so tools are loosely coupled from agent logic. Conversation history threaded through each turn.
Same agentic loop pattern, but with Anthropic's Messages API and Claude's input_schema tool format.
Use Temporal's OpenAI Agents SDK integration. Activities become tools via activity_as_tool - no manual LLM Activity required.
Expose Temporal Workflows as MCP tools using FastMCP. Each tool invocation kicks off a durable Workflow.
Pause an agent on risky actions, wait for human approval via Temporal Signal, then continue or cancel. Wait for hours or days with zero compute cost.
Offload large conversation history to S3 via a PayloadCodec so your Event History stays under the server size limit.
Multi-agent research pipeline: planning, query generation, web search, report synthesis. Survives failures across dozens of LLM calls.
Install Temporal-aware skills and an MCP server into your AI coding assistant.
Hands AI coding agents (Claude Code, Cursor, Codex) expert knowledge of determinism rules, Activity patterns, retry policies, testing, versioning, and the pitfalls Temporal devs hit.
/plugin install temporal@temporal-marketplaceHelps AI agents troubleshoot Temporal Cloud connectivity, authentication, and configuration. Useful when your agent is already writing Cloud-deployed apps.
npx skills add https://github.com/temporalio/skill-temporal-cloudReal-time MCP server compiled from docs, courses, the community forum, and Slack. Your AI assistant queries it for best practices instead of guessing.
claude mcp add --scope user --transport http temporal-docs https://temporal.mcp.kapa.aiExplore architecture, patterns, and lessons for building durable agents.
Foundational walkthrough of building durable AI agents on Temporal - what Temporal gives you for free and how to wire it into your agent loop.
Line-by-line walkthrough of a working durable agent - Activities, Workflows, retries, and how the pieces fit together at runtime.
Multi-agent deep research workflow in Python with OpenAI and Temporal. Planning, query generation, web search, report synthesis - end to end.
New courses, tutorials, and learning resources - straight to your inbox.