corespeed-io/zypher-agent

A minimal yet powerful framework for creating AI agents with full control over tools, providers, and execution flow.

What it solves

Zypher Agent provides a framework for building production-ready AI agents that can be integrated directly into applications. It moves away from rigid, predefined workflows by allowing agents to dynamically decide their next steps based on LLM reasoning.

How it works

The system uses a reactive loop where the agent evaluates the current state and determines the necessary actions. It integrates with multiple LLM providers (like Anthropic Claude and OpenAI GPT) through a unified interface and utilizes the Model Context Protocol (MCP) for server connectivity and OAuth authentication. To ensure reliability, it includes a loop interceptor system for customizing behavior and built-in checkpoint management using a Git-based approach to track and revert changes.

Who it’s for

Developers who want to embed autonomous AI agents into their software applications with production-grade controls like concurrency protection, timeouts, and error handling.

Highlights

Agentic Reasoning: Uses a reactive loop rather than a fixed workflow.
Git-Based Checkpoints: Allows tracking, reviewing, and reverting agent-driven changes.
MCP Support: Native integration with Model Context Protocol servers.
Extensible Tooling: Includes built-in tools for terminal commands, file operations, and search, with support for custom tools.
Production Controls: Features configurable timeouts and concurrency protection.