ready-to-work 0.0.0a2

Architect loop framework for AI-driven development

ready-to-work (rtw)

Plan → Execute → Review loop for AI-driven development. Unlike simple chat, rtw orchestrates an agentic loop where the AI actually executes code changes step-by-step.

Key Features

• Step-by-step execution: Each plan step runs individually with its own timeout, enabling early abort on failures and better error attribution
• Swappable backends: Cursor Agent (default), Codex CLI, Claude Code (stubs ready)
• Persistent state: Interrupt and resume runs, inspect intermediate results in .rtw/runs/
• Lessons learned: Cumulative memory across iterations helps the agent avoid repeating mistakes

Installation

pip install ready-to-work

Or with uv:

uv tool install ready-to-work

Prerequisites: Cursor Agent CLI (authenticated), Python 3.11+.

Usage

rtw run task.md                      # Run architect loop
rtw run task.md --max-iter 5         # Limit iterations
rtw run task.md --model sonnet-4.6   # Override model
rtw run task.md --backend codex      # Use different agent backend
rtw list                             # List runs
rtw resume                           # Resume latest run

Agent Backends

Backend	CLI	Status
cursor	cursor-agent -p	✅ Implemented
codex	codex	🔧 Stub (contributions welcome)
claude	claude	🔧 Stub (contributions welcome)

Development

git clone https://github.com/joey-lou/ready-to-work.git
cd ready-to-work
uv sync
uv run pre-commit install

Run tests and linting:

uv run pytest
uv run ruff check . && uv run ruff format .

Releasing

Version is derived from git tags via hatch-vcs. To release:

git tag v0.3.0 && git push --tags

This triggers the release workflow, which publishes to PyPI and creates a GitHub Release.

License

MIT