Repository navigation

test(evals): mobile behavior eval and a baseline for the iOS hill climb - #25876

Closed

t3dotgg wants to merge 19 commits into

mainfrom

t3/evals-mobile-behavior

Closed

test(evals): mobile behavior eval and a baseline for the iOS hill climb#25876
t3dotgg wants to merge 19 commits into
mainfrom
t3/evals-mobile-behavior

Commits on Oct 27, 2026
- test(mobile): add a behavior eval for the iOS app
  
  t3dotgg
  and
  claude
  committed
  294ad8c
- test(mobile): fix grading gaps from council review
  
  t3dotgg
  and
  claude
  committed
  cbc9794
- test(mobile): stabilize behavior eval flows
  
  t3dotgg
  and
  claude
  committed
  d993cc7
- chore(evals): satisfy lint in mobile behavior eval
  
  t3dotgg
  and
  claude
  committed
  a0473c0
- test(mobile): address second council round on behavior eval
  
  t3dotgg
  and
  claude
  committed
  21b27bd
- test(mobile): grade swipe jank on JS frames, foreground through a link
  
  t3dotgg
  and
  claude
  committed
  a6e8392
- test(mobile): wait for the thread list, not Home, after reload
  
  t3dotgg
  and
  claude
  committed
  d6c55bc
- test(mobile): open the Working shelf when a row is folded in it
  
  t3dotgg
  and
  claude
  committed
  e526ccc
- chore(evals): drop issue number from comment
  
  t3dotgg
  and
  claude
  committed
  0bc6107
- test(mobile): let a new row settle before tapping it
  
  t3dotgg
  and
  claude
  committed
  095e1fb
- test(mobile): start every sample on Home
  
  t3dotgg
  and
  claude
  committed
  1fbed20
- test(mobile): scroll Home to the top before giving up on a row
  
  t3dotgg
  and
  claude
  committed
  49caa91
- test(mobile): label cases from the baseline and update run cost
  
  t3dotgg
  and
  claude
  committed
  91c9113
- test(mobile): wait for the Home link, open the shelf only when collapsed
  
  t3dotgg
  and
  claude
  committed
  9504a63
- test(mobile): archive earlier samples' threads, read the baseline first
  
  t3dotgg
  and
  claude
  committed
  bebea7d
- fix(evals): read the token from stdout, reject unknown case ids, default the forget flag
  
  t3dotgg
  and
  claude
  committed
  eb6892e
- test(mobile): run on the iOS 27.1 Duo simulator
  
  t3dotgg
  and
  claude
  committed
  b81fb00
- test(mobile): wait for the keyboard and give the eval project its own skill
  
  t3dotgg
  and
  claude
  committed
  4910019
- Merge main into t3/evals-mobile-behavior
  t3dotgg
  committed
  2afaf55

Commits

Commits on Oct 27, 2026