usepr/eva

What it solves

EVA is a lightweight AI agent designed to automate technical tasks like writing scripts, creating test cases, executing shell commands, and analyzing data. It aims to provide a high-portability, low-overhead alternative to complex agent frameworks, allowing users to deploy a functional agent as a single Python file.

How it works

EVA operates as a single-file Python script (eva.py) that connects to any LLM supporting the OpenAI API format (including local models via vLLM or Ollama, or cloud APIs like DeepSeek and OpenAI). It maintains session history scoped to the current directory, meaning it remembers context when restarted in the same folder. It also uses a configuration file (EVA.md) to define its persona, settings, and skills.

Who it’s for

It is built for developers and system administrators who need a quick, portable AI assistant for terminal-based automation and shell execution without the need for complex installation processes.

Highlights

• Single-File Architecture: The entire agent is contained in one eva.py file for easy "paste and go" deployment.
• Directory-Scoped Sessions: Automatically persists and resumes conversations based on the working directory.
• Security Review: By default, it requires user confirmation for write/execute commands, though this can be bypassed with a specific flag.
• Local Model Support: Compatible with any OpenAI-compatible API, enabling fully local execution.
• Goal Mode: Includes a -g flag to allow the agent to loop until a specific goal is achieved.