ene-agent 0.1.3

Ene
pip install ene-agent   •   Documentation

An experimental agent harness for personal use.
It's as powerful as other modern agents, but also pythonic and educational for you to understand what happens behind each tool/skill.

Features

• Nothing is Unknown: Explicit context and system prompt, no unexpected memory: You know the agent.
• Skill-first Design: Hierarchical skill loading makes the core small, but capacity large. Memory is also skill.
• Optimized Bundled Skills: Plan, review, clean up, create skills, background processes, pdf-reading, ...
• Personal Skill Library: Use a github repository to easily synchronize your skills.
• Terminal-Native & Web UI: Native terminal experience, while also attached to a modern web UI.

Quick Start

pip install ene-agent

# or from github source
pip install git+https://github.com/ashawkey/ene.git

Configure a model in ~/.ene.yaml, then start chatting:

openai: # openai-compatible
  deepseek: # model alias
    model: deepseek-v4-pro # model name
    base_url: https://api.deepseek.com # base URL
    api_key: ... # your API key

It also supports codex subscription login, please check the documentation for details.

ene [NAME] # start a new, optionally named session
ene resume # resume a previous session (alias: r)

ene models # list available models
ene list # list live sessions (aliases: ls, l)
ene attach [NAME_OR_ID] # attach to a live session (alias: a)
ene kill [NAME_OR_ID] # terminate one or interactively select sessions (alias: k)
ene status # check the status of the .ene folder
ene clean # remove disposable data such as tool results and scratch files
ene clean --history # also remove saved conversation sessions
ene hub # start web UI hub (need to run in background)
ene update # update to the latest source code from github
ene lib # skill library management

WARNING: it has the same permission as the shell user and NO safety guard, use at your own risk.