Download bundle/README.md from earakely-scale/VYMFromRequestComputerEnv-REA: direct link, hf CLI and curl.
- Browser
- Download file 4 kB
-
https://huggingface.co/datasets/earakely-scale/VYMFromRequestComputerEnv-REA/resolve/main/bundle/README.md
- Command line
-
hf download hf://datasets/earakely-scale/VYMFromRequestComputerEnv-REA/bundle/README.md
-
curl -L -o README.md https://huggingface.co/datasets/earakely-scale/VYMFromRequestComputerEnv-REA/resolve/main/bundle/README.md
VYM - View Your Mind computer app
Exported by REA's export-desktop-application from a capture of the vym package
(capture Evidence ev_74658764ab5670c89f9597eb08bca962013d7c0ee3d0247741a4d59e2eb9adbd). It was checked against agentenv-framework
0.9.1431.
envs/vym/is a computer app on theubuntu-24.04base with thereferencedriver, foramd64andarm64. It installsvymfrom the distribution archive, launchessh -c cd "$HOME" 2>/dev/null; exec 'vym' "$HOME"/'Documents/team-offsite.vym'at boot and is ready when a window of classvymappears.- At launch the app opens
~/Documents/team-offsite.vym, the seed's one document. - Owned paths:
~/.config/InSilmaril,~/Documents.data/getreturns their files and a typed JSON view of each;data/resetrestores the seed. - Seed: 2 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. - Neither owned nor seeded, so the app starts without what it saved there: these files or folders, which changed
during the capture,
~/.config/QtProject.conf. To own them, capture again naming them withstate_pathsonstart_build_capture, or--state-pathoncapture-desktop-application. 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/vym/seed/ before publishing the bundle.
Task add-boat-tour-branch
tasks/add-boat-tour-branch.json deploys the app, has the agent work through the computer from the prompt below, and grades
the outcome with artifacts/add-boat-tour-branch-check/verify.py, which reads the app's data/get views and needs
no model. It grades once the app's files have held for 3 s, so a save the app makes on its own within that
time of the agent stopping counts; each result's settled is false when they were still changing 15 s after the
first read. evals/vym.toml names it, so a run of the bundle runs it.
In VYM, add a branch called Sunset boat tour under Agenda in my Team Offsite mind map (Documents/team-offsite.vym) and save it.
Checks, each of which must pass:
new-branch:{"kind":"text_equals","file":"Documents/team-offsite.vym","pointer":"/entries/[name=map.xml]/view/root/children/[name=mapcenter]/children/[name=branch][>heading=Agenda]/children/[name=branch][>heading=Sunset boat tour]/children/[name=heading]","value":"Sunset boat tour"}agenda-has-three:{"kind":"member_count","file":"Documents/team-offsite.vym","pointer":"/entries/[name=map.xml]/view/root/children/[name=mapcenter]/children/[name=branch][>heading=Agenda]/children","matching":"[name=branch]","equals":3}guard-1:{"kind":"text_equals","file":"Documents/team-offsite.vym","pointer":"/entries/[name=map.xml]/view/root/children/[name=mapcenter]/children/[name=branch][@uuid={4cdcd61f-90fb-4965-91a1-849509829d26}]/children/[name=branch][@uuid={8e689d15-a04a-4c60-81ad-6317a3b5874f}]/children/[name=heading]","value":"Strategy workshop"}guard-2:{"kind":"text_equals","file":"Documents/team-offsite.vym","pointer":"/entries/[name=map.xml]/view/root/children/[name=mapcenter]/children/[name=branch][@uuid={4cdcd61f-90fb-4965-91a1-849509829d26}]/children/[name=branch][@uuid={d47c1733-7117-42b7-b45b-9ac244bc3d67}]/children/[name=heading]","value":"Kickoff talk"}
REA kept the task because its checks fail on the seed (new-branch, agenda-has-three) 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-boat-tour-branch --agent claude-code --model MODEL