agent checkpoint
The saved state of a paused agent run, from which it can resume later, for example after a person approves a tool call.
In more detail
Frameworks name it differently: run state in the OpenAI Agents SDK, a checkpointer in LangGraph, sessions in the Claude Agent SDK. Store it safely and validate it on resume, since anything that edits it can change what the agent does next.
Related terms
- human approval — Pausing an agent before a risky action so a person can approve, edit or reject it, then resuming from saved state.
- LangGraph — LangChain's library for building agents and workflows as graphs of steps with saved state.
Taught in
- The Request And The Response — AI Engineering: LLM Apps, RAG & Agents
- Agent SDKs And Human Approval — AI Engineering: LLM Apps, RAG & Agents
- Excessive Agency — AI Engineering: LLM Apps, RAG & Agents