Download bundle/README.md from earakely-scale/OsmoFromRequestComputerEnv-REA: direct link, hf CLI and curl.
- Browser
- Download file 2.76 kB
-
https://huggingface.co/datasets/earakely-scale/OsmoFromRequestComputerEnv-REA/resolve/main/bundle/README.md
- Command line
-
hf download hf://datasets/earakely-scale/OsmoFromRequestComputerEnv-REA/bundle/README.md
-
curl -L -o README.md https://huggingface.co/datasets/earakely-scale/OsmoFromRequestComputerEnv-REA/resolve/main/bundle/README.md
Osmo computer app
Exported by REA's export-desktop-application from a capture of the osmo package
(capture Evidence ev_b5f9fb22877191be9e01ea1c59da700aa9467dda17e331a903ff6ad766776ab0). It was checked against agentenv-framework
0.9.1431.
envs/osmo/is a computer app on theubuntu-24.04base with thereferencedriver, foramd64andarm64. It installsosmofrom the distribution archive, launchessh -c cd "$HOME" 2>/dev/null; exec 'osmo'at boot and is ready when a window of classOsmoappears.- Owned paths:
~/.config/osmo,~/.local/share/osmo.data/getreturns their files and a typed JSON view of each;data/resetrestores the seed. - Seed: 7 file(s) made for the task, as saved at checkpoint
seed. Text files and dconf keys spell the capture host's name aslocalhost,user@hostasuser@localhostand the home as/home/user, which the computer respells to its own at boot. tasks/smoke.jsondeploys the app.
Run the task with agent-env run <this folder>, or only the deploy with --task smoke.
The seed was made for this task, not taken from a user's own data; read envs/osmo/seed/ before publishing the bundle.
Task add-contact
tasks/add-contact.json deploys the app, has the agent work through the computer from the prompt below, and grades
the outcome with artifacts/add-contact-check/verify.py, which reads the app's data/get views and needs
no model. evals/osmo.toml names it, so a run of the bundle runs it.
In Osmo, add a new contact to my address book: Delphine Moreau, cell phone 555-0163, e-mail delphine.moreau@example.com.
Checks, each of which must pass:
name:{"kind":"text_contains","file":".local/share/osmo/contacts_records.xml","pointer":"/root/children/[name=contacts_records]","value":"DelphineMoreau"}phone:{"kind":"text_contains","file":".local/share/osmo/contacts_records.xml","pointer":"/root/children/[name=contacts_records]","value":"555-0163"}email:{"kind":"text_contains","file":".local/share/osmo/contacts_records.xml","pointer":"/root/children/[name=contacts_records]","value":"delphine.moreau@example.com"}kept:{"kind":"text_contains","file":".local/share/osmo/contacts_records.xml","pointer":"/root/children/[name=contacts_records]","value":"hannah.okafor@example.com"}
REA kept the task because its checks fail on the seed (name, phone, email) 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 add-contact --agent claude-code --model MODEL