agentsop-langgraph

What problem does it solve?

It solves the uncertainty coders face when designing production-grade LangGraph agent systems, especially around when to use specific APIs, how to structure state and reducers, and how to avoid common runtime failure modes like invalid parallel updates or runaway loops.

Core Features & Use Cases

• Decision Protocol for Architecture: Choose between prebuilt ReAct, functional API, and full StateGraph based on statefulness, topology, and production needs.
• Operational SOP for Reliability: Design typed state, add mandatory reducers for parallel writes, configure checkpointers for durability, and implement safe human-in-the-loop using interrupt/resume.
• Production Debugging Playbook: Handle known LangGraph failure modes (e.g., recursion_limit misuse), bound cyclic workflows with explicit retry counters, and use time-travel via checkpoints to reproduce and fork failed runs.
• Use Case: When migrating a fragile tool-calling LangChain loop into a durable, restart-safe workflow with HITL approvals (e.g., reviewing high-blast-radius tool actions), this Skill provides the exact gates, topology choices, and safety rules.

Quick Start

Tell a coder-agent: "Use the agentsop-langgraph Skill to guide the architecture and SOP for my LangGraph workflow, including state schema, reducers, checkpoint choice, HITL interrupt placement, and bounded failure handling."