🌉ODSC AI West 2025Official Partner & Exhibitor
San FranciscoOct 28-30
Our ODSC Story
New Release
Open Source

Introducing Agenspy

Protocol-First Agent Framework Built on DSPy

A next-generation framework that brings structure, interoperability, and production-readiness to AI agents with first-class protocol support.

While DSPy gives you the power to define and optimize declarative agent programs, we felt something was missing: a protocol layer.

The Questions We Asked:

  • • How do agents interact across services?
  • • How do we define multi-agent collaboration?
  • • What if agents need to talk to external tools in a structured way?

Agenspy answers these questions with a protocol-first approach to AI agents, bringing structure and interoperability to the agent ecosystem.

Get Started in Seconds

install.shInstallation
bash
pip install "agenspy[mcp]"
terminalTry it now
bash
agenspy demo github-pr

Quick Example

quick_example.pyProtocol-Aware GitHub PR Review Agent
python
from agenspy import create_mcp_pr_review_agent
import dspy

# Configure DSPy
lm = dspy.LM('openai/gpt-4o-mini')
dspy.configure(lm=lm)

# Create protocol-aware agent
agent = create_mcp_pr_review_agent("mcp://github-server:8080")

# Use the agent
result = agent(
    pr_url="https://github.com/org/repo/pull/123", 
    review_focus="security"
)

print(result.review_comment)

Features at a Glance

Protocol-First Architecture

Built-in support for MCP, A2A, and future protocols with automatic connection management

DSPy Integration

Native integration with DSPy for reasoning, prediction, and optimization

Full-featured CLI

Complete command-line interface for building, running, and testing agents

Future-Proof Protocols

Extensible architecture for adding new protocols like WebSocket, gRPC, etc.

Multi-Protocol Support

Support for MCP, A2A, and extensibility for future protocols

Full-Stack Agents

Python & JavaScript protocol servers for complete agent ecosystems

Architecture Overview

Protocol-First Architecture

How Agenspy brings protocol intelligence to DSPy agents

DSPy Agent Layer

ChainOfThought
ReAct Patterns
Optimized Prompts

Protocol Intelligence

Auto Connection
Session Management
Protocol Extensibility
MCP Support

Model Context Protocol for tool integration

A2A Protocol

Agent-to-Agent communication

Future Protocols

WebSocket, gRPC, and more

Examples to Explore

basic_mcp_demo.py

View

Hello World Agent

Simple MCP agent implementation with DSPy integration

github_pr_review.py

View

GitHub PR Review

Automated pull request review with security focus

multi_protocol_demo.py

View

Multi-Protocol Demo

Experimental protocol stack demonstration

python_server_demo.py

View

Python MCP Server

Local MCP agent server implementation

What's Next for Agenspy?

Agenspy is just getting started. Our vision is to make it a first-class DSPy protocol extension, listed in Google's A2A Agent Directory, and fully ready for multi-agent production environments.

Protocol Expansions

  • • WebSocket & gRPC-based layers
  • • Real-time collaboration protocols
  • • Agent network discovery

Tooling & Infrastructure

  • • Dashboards & orchestration
  • • Cloud-native deployment
  • • Advanced debugging tools

Ready to Build Protocol-Native Agents?

Join the next generation of AI agent development with protocol-first architecture.

PyPI Package
Open Source
Full CLI