What problem does it solve? LangGraph agents that need human approval, review, or input mid-execution require correct use of interrupt(), Command(resume=...), checkpointers, and thread IDs, and mistakes cause duplicate side effects, stuck graphs, or lost state. ## Core Features & Use Cases - Interrupt and Resume Patterns: Pause graph execution with interrupt() and resume with Command(resume=...) using a checkpointer and thread ID. - Approval and Validation Workflows: Route based on human decisions, validate human input in re-prompt loops, and resume multiple parallel interrupts via ID-mapped resume values. - Idempotency and Error Guidance: Prevent duplicate side effects on node re-execution and avoid antipatterns like passing Command(update=...) as invoke input. - Use Case: When building an email-drafting agent that must pause for a human to approve or edit the draft before sending, use this Skill to wire the interrupt, route on the decision, and avoid duplicate audit-log writes on resume. ## Quick Start Ask the AI to add a human approval step to your LangGraph agent that pauses with interrupt() and resumes with the reviewer's decision.