earakely-scale/LibreOfficeCalcFromRequestComputerEnv-REA

LibreOffice Calc: a computer-use environment built from a one-line request

The real LibreOffice Calc 24.2 spreadsheet 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 36 minutes, with only REA's build skill and MCP tools; How it was made has the request and the steps. LibreOffice Calc opens a small quarterly sales spreadsheet, Documents/q1-sales.ods, at launch: January, February and March sales in US dollars, and a Q1 Total row that sums them. The agent has to correct February's figure from 21,300 to 23,100 and save, leaving the other months alone. 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

February's sales figure in my quarterly sales spreadsheet, Documents/q1-sales.ods, is wrong: it should be 23,100. Fix it and save the spreadsheet.

bundle/artifacts/fix-february-sales-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
february	text_contains	Documents/q1-sales.ods
sheet	text_equals	Documents/q1-sales.ods

An .ods file is a zip, and REA reads content.xml inside it. The checks find the sheet by its name, Q1 Sales, never by position, and read the sheet's whole text, its cells' text run together: it must contain "February23100March" and equal "MonthSales (USD)January18450February23100March19875Q1 Total61425" exactly. LibreOffice saves a formula's result alongside the formula, so the exact match also proves that the other months are untouched and that the Q1 Total recalculated to 61,425. REA's draft suggested eight checks, all on things any editing session changes, such as the cursor position, the document's editing time and the time field in the page header; Claude Code discarded them and wrote these.

Results

3 of 3 runs passed, about $0.58 per run.

Run	Outcome	Turns	Cost	Video
1	passed	11	$1.02	video
2	passed	11	$0.25	video
3	passed	10	$0.48	video

Run it

You need Docker. agentenv-framework builds the computer: Ubuntu 24.04 with the app installed from Ubuntu's archive (libreoffice-calc 24.2.7).

pip install agentenv-framework==0.9.1431
hf download earakely-scale/LibreOfficeCalcFromRequestComputerEnv-REA --repo-type dataset --local-dir calc-request
agent-env run calc-request/bundle --task fix-february-sales --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 LibreOffice Calc environment where the agent fixes one month's figure in a small quarterly sales spreadsheet and saves it.

It asked only the skill's upfront questions, wrote no app-specific scripts, and finished in about 36 minutes.

It took a second capture. The first export seeded LibreOffice's profile lock file, ~/.config/libreoffice/4/.lock, so in a fresh computer LibreOffice found its profile locked and would not start. Claude Code captured again, deleting the lock file in the file manager before the seed was saved and closing the spreadsheet before the end. The steps were REA's usual ones:

1. Capture. REA launched LibreOffice Calc 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: February is 21300
after_edit	pass
near_miss	fail, as expected: February was typed as 23010
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 checks read the cells' text as LibreOffice saves it, not their numeric values, so February must be saved as 23100. Shown with a thousands separator, as "23,100", it fails though its value is right.
• The sheet check compares the whole sheet's text exactly, so anything else typed on the sheet, even in a spare cell, fails it.
• The seed holds LibreOffice's user profile as the capture left it, including its icon cache (~/.config/libreoffice/4/cache, 558 files). LibreOffice started cleanly from it in the witness and in every run.
• The computer installs whatever libreoffice-calc Ubuntu 24.04's archive holds at build time (24.2.7 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. LibreOffice Calc itself is not included: the computer installs it from Ubuntu's archive (libreoffice-calc 24.2.7) at build time, under its own licence (MPL-2.0).