ml-intern-codex

ml-intern-codex is a local-first ML engineering terminal app built on top of your installed codex.

• full-screen Rust TUI
• local wrapper app-server
• skill-first ML workflows
• local artifact persistence for reports, audits, and job snapshots

Requirements

• codex on PATH
• a working local Codex auth/config under ~/.codex
• curl
• tar
• python3

Quick check:

codex --version

One-line install

Latest release install:

curl -fsSL \
  https://raw.githubusercontent.com/BetterAndBetterII/ml-intern-codex/main/scripts/install-remote.sh \
  | bash

This downloads the latest GitHub release for your platform and installs:

• release files into ~/.local/share/ml-intern-codex
• launchers into ~/.local/bin

Local install from clone

If you already cloned the repo:

./scripts/install.sh

If ml-intern is still not found after install, open a new shell or run:

. "$HOME/.cargo/env"

Quick Start

Start the TUI:

ml-intern

Useful first commands inside the app:

• $ — open the skill picker
• /skills — list skills
• /threads — resume a thread
• /artifacts — browse saved artifacts
• /help — show keybindings

Start the local app-server directly:

ml-intern-app-server

Development

Run the workspace tests:

cargo test --workspace

See docs/IMPLEMENTATION_STATUS.md for implementation notes and current validation status.