earakely-scale/GladeFromRequestComputerEnv-REA

Glade: a computer-use environment built from a one-line request

The real Glade 3.40 GTK interface designer 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.

Claude Code built this environment from a one-line request in about 19 minutes, with only REA's build skill and MCP tools; How it was made has the request and the steps. Glade opens a small dialog design, Documents/feedback.glade, at launch: a Send Feedback dialog with an intro label and two buttons, Cancel and Submit. The agent has to change the Submit button's label to Send Feedback and save. It sees only the screen and acts only with the mouse and keyboard.

Watch

• Agent runs (Claude Code with Claude Sonnet 5.5): run 1, run 2, run 3
• Witness, which proves the task can be done through the computer: witness.mp4

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

The task

In Glade, change the label of the Submit button in the feedback dialog design in Documents/feedback.glade to Send Feedback and save it.

bundle/artifacts/rename-submit-button-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
submit-label	text_equals	Documents/feedback.glade
cancel-kept	text_equals	Documents/feedback.glade
two-buttons	member_count	Documents/feedback.glade

A .glade file is GtkBuilder XML. The buttons sit in the dialog's action area, each in a child element whose widget id is held one level down, where no selector can name it, so the checks find each button by its packing properties run together ([>packing=TrueTrue1]: expand, fill and position 1 for Submit). Submit's label must read "Send Feedback" exactly, Cancel's must still read "Cancel", and the action area must hold two buttons (member_count). REA's draft suggested the label check, kept exactly as written, and a Cancel guard whose selector was built from the very text it checked; Claude Code rebuilt that guard on the packing and added the count.

Results

3 of 3 runs passed, about $0.84 per run.

Run	Outcome	Turns	Cost	Video
1	passed	16	$0.94	video
2	passed	16	$0.87	video
3	passed	17	$0.70	video

Run it

You need Docker. agentenv-framework builds the computer: Ubuntu 24.04 with the app installed from Ubuntu's archive (glade 3.40.0, with librsvg2-common).

pip install agentenv-framework==0.9.1431
hf download earakely-scale/GladeFromRequestComputerEnv-REA --repo-type dataset --local-dir glade-request
agent-env run glade-request/bundle --task rename-submit-button --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

Claude Code built this environment from a one-line request, given only REA's build skill and REA's MCP tools:

Build me a Glade environment where the agent changes a button's label in a small dialog design and saves it.

It asked only the skill's upfront questions, wrote no app-specific scripts, and finished in about 19 minutes. The steps were REA's usual ones:

1. Capture. REA launched Glade on a Linux desktop and recorded its package, windows and opened files while the edit was made. 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. In a fresh computer, the edit was made through computer actions alone, and the checks were read at four stages. They held on the first try:

Stage	Checks
seed	fail, as expected: the button reads Submit; the kept checks pass
after_edit	pass
near_miss	fail, as expected: the label was retyped as "Send Feedbak"
after_reset	fail, as expected: data/reset restored the seed

The agent runs came after the build: Claude Code with Claude Sonnet 5.5 (--model anthropic/claude-sonnet-5-5 --no-emulate, on an amd64 host) solved the task 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 buttons are found by their packing, not by their ids: moving Submit to another position in the button box, or changing its expand or fill packing, makes the checks miss it even when its label is right.
• The bundle installs librsvg2-common, the SVG image loader Glade's icons need, beside the app as an extra package.
• The computer installs whatever glade Ubuntu 24.04's archive holds at build time (3.40.0 for these runs); it is not pinned.
• Three runs is a small sample: the solve rate shows the task is solvable, not how hard it is.

Licences

The bundle code, the task and the synthetic seed files are Apache-2.0. Glade itself is not included: the computer installs it from Ubuntu's archive (glade 3.40.0, with librsvg2-common) at build time, under its own licence (LGPL-2.1-or-later and GPL-2.0-or-later).