Termux Codex Helper

A small Termux-first CLI that wraps the OpenAI API for safe terminal workflows:

• ask "question"
• explain path/to/file
• review path/to/file
• cmd "goal"
• ask -f path/to/file "question about this file"
• fix path/to/file "instruction"
• patch path/to/file "instruction" with explicit confirmation before apply

The tool is intentionally conservative:

• read-only by default
• patch generation is opt-in
• file edits require confirmation
• obvious secrets are masked before content is sent to the API

Requirements

• Termux
• bash
• curl
• jq
• patch for applying diffs

Install the missing packages if needed:

pkg install curl jq patch

Setup

Copy .env.example to .env and set your API key:

cp .env.example .env

Or export variables in your shell:

export OPENAI_API_KEY="..."
export OPENAI_MODEL="gpt-5-mini"

Install

Install into ~/.local/bin and add a stable ask launcher:

./install.sh

If ~/.local/bin is not already on your PATH, add this to ~/.bashrc:

export PATH="$HOME/.local/bin:$PATH"

Then reload your shell:

source ~/.bashrc

Usage

Ask

ask "Why is my shell script hanging?"
ask -f lib/watchdog.sh "Explain the restart flow"
cat script.sh | ask "Review this for bugs"

Explain

termux-codex explain lib/watchdog.sh

Review

termux-codex review lib/watchdog.sh

Command Suggestions

termux-codex cmd "find which process is listening on port 18789 in Termux"

Fix

termux-codex fix lib/watchdog.sh "Add a startup grace period before health checks"

Patch

termux-codex patch lib/watchdog.sh "Add a startup grace period before health checks"

The command will:

1. Read the target file
2. Ask the model for a unified diff only
3. Show the diff
4. Run patch --dry-run
5. Ask whether to apply it

Environment Variables

• OPENAI_API_KEY required API key
• OPENAI_MODEL optional model override, default gpt-5-mini
• OPENAI_BASE_URL optional API base URL, default https://api.openai.com/v1

Notes

• The patch workflow is single-file by design.
• The helper masks common KEY=, TOKEN=, SECRET= patterns before sending content.
• For large files, trim the file manually before sending if you want lower token usage.