What is AgentMesh?
AgentMesh is a Go framework for building sophisticated multi-agent AI systems powered by Pregel-style bulk-synchronous parallel (BSP) graph processing. It provides production-grade orchestration for LLM-powered workflows with deterministic execution and enterprise observability.
Graph-native Architecture
Build agents as directed graphs where nodes execute in parallel supersteps with deterministic ordering.
Strongly Typed Tools
Register functions as tools with automatic JSON Schema generation and compile-time type safety.
Streaming Execution
Monitor graph execution in real-time with event streams for responsive user experiences.
Enterprise Ready
Built-in checkpointing, time-travel debugging, OpenTelemetry integration, and production reliability.
Core capabilities
Pregel Graph Execution
Bulk-synchronous parallel processing enables efficient multi-agent coordination with deterministic superstep ordering.
LLM Integration
First-class support for OpenAI, Anthropic, Amazon Bedrock, Google Gemini, and custom model providers.
Tool Orchestration
Type-safe function calling with automatic JSON schema generation, parallel execution, and robust error handling.
State Management
Versioned state store with channel-based updates, automatic checkpointing, and time-travel debugging.
Conditional Routing
Dynamic flow control based on agent outputs, enabling complex decision trees and multi-agent collaboration.
Observability
Built-in OpenTelemetry metrics, distributed tracing, and structured logging for production monitoring.
Quick examples
Explore hands-on examples to see AgentMesh in action:
Explore the documentation
🚀 Getting Started
Install the module, run your first agent, and explore example workflows.
🏗️ Architecture
Understand the Pregel BSP model, graph builder pattern, and state management.
🤖 Agents Guide
Build ReAct agents, RAG agents, and custom graph-based workflows.
🛠️ Tools Guide
Create function tools with automatic schema generation and integrate external capabilities.
🧠 Models Guide
Connect OpenAI, Anthropic, Bedrock, Gemini, or custom LLM providers with routing.
🔭 Observability
Configure OpenTelemetry metrics, tracing, and structured logging.