langgraph-patterns

Orchestrate Gemini-Harness agent registries with six LangGraph IPC patterns.

7|Updated Apr 19, 2026
One-click install
npx skills add https://github.com/namojo/gemini-harness --skill langgraph-patterns-namojo
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: langgraph-patterns
Source: https://github.com/namojo/gemini-harness/tree/main/.claude/skills/langgraph-patterns
Command: npx skills add https://github.com/namojo/gemini-harness --skill langgraph-patterns-namojo

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

LangGraph patterns standardize how Gemini-Harness designs and runs a scalable, meta-agent based architecture. They provide a compact, fixed graph (Manager + Worker + Registry) to express run-time agent registries and dispatch logic on a single worker node, enabling dynamic agent creation and inter-agent communication.

Core Features & Use Cases

  • Manager + Worker + Registry orchestration on a minimal graph to route tasks and manage state.
  • Implements six architectural patterns: pipeline, fan_out_fan_in, expert_pool, producer_reviewer, supervisor, and hierarchical, plus support for complex composites.
  • Self-critique loop and LangGraph adapter compatibility to ensure stable runtimes and forward compatibility.

Quick Start

Initialize the graph with Manager+Worker+Registry and invoke the runtime to start processing the registry.

Frequently Asked Questions about langgraph-patterns

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I orchestrate dynamic agent registries in LangGraph?

You can orchestrate dynamic agent registries in LangGraph using a compact Manager, Worker, and Registry StateGraph to route tasks, manage state, and dispatch runtime agents on a single worker node. This enables dynamic agent creation and inter-agent communication.

What IPC patterns can I use to route tasks for multi-agent workflows?

You can use six architectural IPC patterns to route tasks for multi-agent workflows: pipeline, fan_out_fan_in, expert_pool, producer_reviewer, supervisor, and hierarchical. These patterns support complex composites for scalable meta-agent based architectures.

How do I scale Gemini teams with a fixed StateGraph runtime?

You can scale Gemini teams with a fixed StateGraph runtime by applying a minimal API surface that restricts runtime changes to the registry via reducer-based updates. This ensures predictable messaging and stable runtimes while managing dynamic agent dispatch.

Does LangGraph support self-critique loops for agent runtimes?

Yes, LangGraph supports self-critique loops for agent runtimes by enforcing a stable StateGraph with a minimal API surface. This ensures predictable messaging and adapter compatibility while restricting runtime changes to the registry via reducer-based updates.

What is the best way to manage state changes in a multi-agent graph?

The best way to manage state changes in a multi-agent graph is to restrict runtime modifications to the registry via reducer-based updates. This approach maintains a stable StateGraph and predictable messaging while allowing dynamic agent creation.