bundle/README.md

KiCad Schematic Editor (Standalone) computer app

Exported by REA's export-desktop-application from a capture of the kicad package (capture Evidence ev_586ce4682a16a7015d3ed5fc762a91735214708df4fb45dac25b061f37c64363). It was checked against agentenv-framework 0.9.1431.

• envs/eeschema/ is a computer app on the ubuntu-24.04 base with the reference driver, for amd64 and arm64. It installs kicad from the distribution archive, launches sh -c exec 'eeschema' "$HOME"/'Documents/led/led.kicad_sch' at boot and is ready when a window of class eeschema appears.
• Owned paths: ~/.config/kicad, ~/Documents/led. data/get returns their files and a typed JSON view of each; data/reset restores the seed.
• Seed: 6 file(s) from the user's home before the capture, with the captured home path respelled in text files.
• 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 r1-220

tasks/r1-220.json deploys the app, has the agent work through the computer from the prompt below, and grades the outcome with artifacts/r1-220-check/verify.py, which reads the app's data/get views and needs no model. evals/eeschema.toml names it, so a run of the bundle runs it.

In the USB LED indicator schematic open in KiCad's Schematic Editor, change resistor R1's value from 330 to 220 (ohms), then save the schematic.

Checks, each of which must pass:

• r1-is-220: {"kind":"line_count","file":"Documents/led/led.kicad_sch","contains":"(property \"Value\" \"220\"","equals":1}
• no-330: {"kind":"line_count","file":"Documents/led/led.kicad_sch","contains":"(property \"Value\" \"330\"","equals":0}
• r1-kept: {"kind":"line_count","file":"Documents/led/led.kicad_sch","contains":"(property \"Reference\" \"R1\"","equals":1}
• led-kept: {"kind":"line_count","file":"Documents/led/led.kicad_sch","contains":"(property \"Value\" \"LED_Green\"","equals":1}

REA kept the task because its checks fail on the seed (r1-is-220, no-330) 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 r1-220 --agent claude-code --model MODEL