bundle/README.md

AbiWord computer app

Exported by REA's export-desktop-application from a capture of the abiword package (capture Evidence ev_d5fbe84e3808b08f5b92060897d55564d9ffa4c6da301b681cfec8eb78ed0990). It was checked against agentenv-framework 0.9.1431.

• envs/abiword/ is a computer app on the ubuntu-24.04 base with the reference driver, for amd64 and arm64. It installs abiword from the distribution archive, launches sh -c cd "$HOME" 2>/dev/null; exec 'abiword' at boot and is ready when a window of class Abiword appears.
• Owned paths: ~/.config/abiword, ~/Documents. data/get returns their files and a typed JSON view of each; data/reset restores the seed.
• Seed: 2 file(s) from the user's home at checkpoint seed. Text files and dconf keys spell the capture host's name as localhost, user@host as user@localhost and the home as /home/user, which the computer respells to its own at boot.
• tasks/smoke.json deploys the app.

Run the task with agent-env run <this folder>, or only the deploy with --task smoke.

The seed is synthetic test data made for this showcase; it holds no one's real files.

Task change-meeting-time

tasks/change-meeting-time.json deploys the app, has the agent work through the computer from the prompt below, and grades the outcome with artifacts/change-meeting-time-check/verify.py, which reads the app's data/get views and needs no model. evals/abiword.toml names it, so a run of the bundle runs it.

The book sale planning meeting has moved. Change the meeting time in the memo Documents/book-sale-memo.abw to 2:30 PM and save it.

Checks, each of which must pass:

• meeting-time: {"kind":"text_equals","file":"Documents/book-sale-memo.abw","pointer":"/root/children/[name=section]/children/[name=p][@xid=6]","value":"The spring book sale planning meeting will be held on Thursday, 12 February at 2:30 PM in the Community Room on the second floor.","normalize_whitespace":true}
• agenda-deadline-kept: {"kind":"text_contains","file":"Documents/book-sale-memo.abw","pointer":"/root/children/[name=section]/children/[name=p][@xid=7]","value":"by 10:00 AM on Wednesday","normalize_whitespace":true}

REA kept the task because its checks fail on the seed (meeting-time) and all pass on the captured end state.

To measure a solve rate, run the task k times (3 by default) with the agent and model to measure, and count the runs whose summary line says passed; a run that says failed did not finish and is run again rather than counted:

agent-env run <this folder> --task change-meeting-time --agent claude-code --model MODEL