earakely-scale's picture
REA computer-use environment, v0.1.0
4e8f4e3 verified
|
Raw History Blame Contribute Delete
3.87 kB
metadata
license: apache-2.0
pretty_name: LibreCAD computer-use environment (REA)
tags:
  - computer-use
  - cua
  - gui-agent
  - rl-environment
  - agentenv
  - desktop
  - linux
  - rea
  - cad

LibreCAD: a computer-use environment

An agent solving the task in LibreCAD

The real LibreCAD 2.2 2D CAD running on an Ubuntu 24.04 desktop, as an agentenv computer environment. An agent completes the task only through screenshots, mouse and keyboard, and a verifier grades the saved files with no model.

LibreCAD opens a mounting plate drawing with one centre hole. The agent has to add four more mounting holes as circles and save the DXF, working only with LibreCAD's tools through the screen, mouse and keyboard.

Watch

Each video is the computer's own screen recording at 10 frames per second.

The task

Add four more mounting holes to the plate drawing in LibreCAD and save it.

bundle/artifacts/add-holes-check/verify.py reads the app's files through the computer's data/get view and passes only if every check holds:

Check Kind File
nine-circles line_count Documents/plate.dxf

The checks read the saved DXF: it now holds 9 circle entities (the plate's 5 starting circles plus the 4 new holes).

Results

3 of 3 runs passed, about $0.49 per run.

Run Outcome Cost Video
1 passed $0.81 video
2 passed $0.32 video
3 passed $0.34 video

Run it

You need Docker. agentenv-framework builds the computer: Ubuntu 24.04 with the app installed from Ubuntu's archive (librecad 2.2.0.2-1build3).

pip install agentenv-framework==0.9.1431
hf download earakely-scale/LibreCADComputerEnv-REA --repo-type dataset --local-dir librecad
agent-env run librecad/bundle --task add-holes --agent claude-code --model <model>

The agent needs a model provider configured for agentenv. --task smoke only boots the computer, with no agent. bundle/README.md describes the bundle in detail.

How it was made

REA built this environment from a real session with the app:

  1. Capture. REA launched LibreCAD on a Linux desktop and recorded its package, windows and opened files while a scripted user made the edit. It also recorded the app's state folders before and after.
  2. Export. REA turned the capture into an agentenv computer_app. It installs the app, seeds the starting files, starts the app, and waits for its main window. It also serves typed views of the app's files for verifiers.
  3. Task. The task's checks were validated against the capture: they fail on the starting state and pass on the captured end state.
  4. Witness and runs. Scripted computer actions solved the task in a kept computer, and the verifier failed before and passed after. Then an agent solved it 3 times, each run recorded.

What's inside

bundle/      the agentenv bundle: envs/ (app module, seed files), tasks/, artifacts/ (verifier), evals/
media/       witness and agent-run recordings (MP4) with short GIF previews
results.json the runs above, with each check's result

Limits

  • The check counts circles; it does not check where they are or their size.
  • The computer uses the librecad version in Ubuntu 24.04's archive at build time; it is not pinned.
  • Three runs is a small sample.

Licences

The bundle code, the task and the synthetic seed files are Apache-2.0. LibreCAD itself is not included: the computer installs it from Ubuntu's archive (librecad 2.2.0.2-1build3) at build time, under its own licence (GPL-2.0).