📌Overview

Purpose:
To provide a simple, powerful toolkit for building, running, and orchestrating LLM-powered agents using a declarative approach suitable for programmable and auditable automation.

Overview:
Nerve is an Agent Development Kit (ADK) that enables technical users to develop, deploy, and evaluate agents based on large language models (LLMs) with minimal code. Leveraging YAML for agent definitions and a command-line interface (CLI), Nerve focuses on creating reproducible, transparent, and extensible automations. Its platform-agnostic design supports various LLM providers and promotes collaborative and orchestrated agent workflows.

Key Features:

• Declarative YAML Agent Definitions:
  Agents are described using clean, auditable YAML files, consolidating prompts, tasks, tools, and variables for easy maintenance and sharing.

• Extensible Tools Integration:
  Support for shell commands, Python functions, and remote tools, all with strong typing and annotations, allows for rich and customizable agent capabilities.

• MCP Server/Client Support:
  Enables both defining and running MCP servers and clients natively in YAML, supporting advanced agent orchestration and collaborative agent teams.

• Built-in Evaluation Mode:
  Provides robust benchmarking with YAML, Parquet, or folder-based test cases, supporting reproducible testing and detailed progress tracking.

• Composable Workflows:
  Allows agents to be assembled into linear pipelines, enabling multi-step automations with shared context between steps.

• LLM-Agnostic Architecture:
  Built on LiteLLM, supporting interoperability with a wide range of LLMs (OpenAI, Anthropic, Ollama, and more) for flexible model selection.

---

nerve

The Simple Agent Development Kit

Nerve is a simple yet powerful Agent Development Kit (ADK) to build, run, evaluate, and orchestrate LLM-based agents using just YAML and a CLI. Designed for technical users, it enables programmable, auditable, and reproducible automation with large language models.

Key Features

• Declarative Agents
  Define agents using a clean YAML format. Specify system prompt, task, tools, and variables in one file.

• Built-in Tools & Extensibility
  Use shell commands, Python functions, or remote tools to power your agents. Tools are fully typed and annotated.

• Native MCP Support (Client & Server)
  Define MCP servers in YAML, and operate as both client and server, enabling agent teams and deep orchestration.

• Evaluation Mode
  Benchmark your agents with YAML, Parquet, or folder-based test cases. Run reproducible tests, log structured outputs, and track regression or progress.

• Workflows
  Compose agents into simple, linear pipelines for multi-step automations with shared context.

• LLM-Agnostic
  Built on LiteLLM, Nerve supports OpenAI, Anthropic, Ollama, and dozens more providers. Switch models with a single line.

Quick Start

Install and use Nerve via the CLI:

pip install nerve-adk

nerve install evilsocket/changelog

nerve create new-agent

nerve run new-agent

For more information, see the documentation and examples.

Contributing

Contributions are welcome! See our contributing guidelines and join the Discord community for help and discussion.

License

Nerve is released under the GPL 3 license.