Neura Market
<div align="center"><img src="https://raw.githubusercontent.com/evilsocket/search/refs/heads/main/logo.png" height="20"/></div> logo

<div align="center"><img src="https://raw.githubusercontent.com/evilsocket/search/refs/heads/main/logo.png" height="20"/></div>

Free

[Nerve](https://github.com/evilsocket/nerve)

FreeFree tier
Type
Open Source
Visit Website GitHub (40 stars)

About <div align="center"><img src="https://raw.githubusercontent.com/evilsocket/search/refs/heads/main/logo.png" height="20"/></div>

Nerve is a simple yet powerful Agent Development Kit (ADK) designed for building, running, evaluating, and orchestrating LLM-based agents using YAML and a CLI. It targets technical users who need programmable, auditable, and reproducible automation powered by large language models. Key capabilities include declarative agent definitions in YAML, extensibility via shell commands, Python functions, or remote tools, native MCP (Model Context Protocol) support enabling agents to act as both client and server, an evaluation mode for benchmarking agents with YAML, Parquet, or folder-based test cases, and workflow composition for multi-step automations with shared context. Nerve is LLM-agnostic, built on LiteLLM, and supports OpenAI, Anthropic, Ollama, and many other models. Installation is via pip, and agents can be installed from GitHub repos or created interactively. The project is released under the GPL 3 license and the repository is currently archived and read-only as of February 2026.

Key Features

Declarative Agents: Define agents using a clean YAML format with system prompt, task, tools, and variables.
Built-in Tools Extensibility: Use shell commands, Python functions, or remote tools; fully typed and annotated.
Native MCP Support (Client & Server): First framework to define MCP servers in YAML and act as both client and server for agent teams and orchestration.
Evaluation Mode: Benchmark agents with YAML, Parquet, or folder-based test cases; run reproducible tests and track regression.
Workflows: Compose agents into simple linear pipelines for multi-step automations with shared context.
LLM-Agnostic: Built on LiteLLM, supports OpenAI, Anthropic, Ollama, and dozens more; switch models in one line.

Pros & Cons

Pros
  • Declarative YAML configuration makes agent definition simple and auditable
  • Extensible tool system supports shell, Python, and remote tools
  • First-class MCP support enables advanced agent coordination
  • LLM-agnostic design allows easy switching between models
  • Includes evaluation mode for benchmarking and regression tracking
  • Free and open source under GPL 3 license
Cons
  • Repository is archived and read-only as of February 2026, indicating no active maintenance
  • Requires technical knowledge of YAML and CLI; not beginner-friendly
  • Limited to CLI usage without a graphical interface

Best For

Building and running LLM-based agents for automation tasksOrchestrating multi-agent teams with MCP supportEvaluating and benchmarking agent performance with structured test casesCreating reproducible, auditable automation pipelinesRapid prototyping of AI agents using YAML declarations

FAQ

What is Nerve?
Nerve is a simple Agent Development Kit (ADK) for building, running, evaluating, and orchestrating LLM-based agents using YAML and a CLI.
How do I install Nerve?
Install via pip with the command: pip install nerve-adk
What models are supported?
Nerve is built on LiteLLM and supports OpenAI, Anthropic, Ollama, and dozens of other models. You can switch models with a single line change.
Does Nerve support MCP?
Yes, Nerve is the first framework to let you define MCP servers in YAML and act as both client and server, enabling agent teams and deep orchestration.
Is Nerve free?
Yes, Nerve is open source and released under the GPL 3 license. There is no paid version.