Repository navigation

fix(evals): simulated delegate_task results carry every field T3 returns - #28824

Closed

only21mil wants to merge 9 commits into

pingdotgg:mainfrom

only21mil:fix/audit-m265

Closed

fix(evals): simulated delegate_task results carry every field T3 returns#28824
only21mil wants to merge 9 commits into
pingdotgg:mainfrom
only21mil:fix/audit-m265

Commits on Nov 1, 2026
- fix(evals): delegated-task eval results match the MCP contract
  
  only21mil
  and
  claude
  committed
  7657268
- refactor(evals): reuse the contract's terminal statuses and keep replays to the contract
  
  only21mil
  and
  claude
  committed
  8fb07b7
Commits on Nov 2, 2026
- fix(evals): merge main preserving task and thread simulations
  maria-rcks
  committed
  ac7c4f8
Commits on Nov 10, 2026
- Merge main into fix/audit-m265
  
  only21mil
  and
  claude
  committed
  12dc8ce
- fix(evals): simulated delegated-task results include executedModel
  
  only21mil
  and
  claude
  committed
  6aa377f
- test(evals): delegated-task shape test requires executedModel
  
  only21mil
  and
  claude
  committed
  468d7b8
Commits on Nov 12, 2026
- Merge main into fix/audit-m265
  
  only21mil
  and
  claude
  committed
  be11f65
- fix(evals): later task reads keep a finished task's result
  
  only21mil
  and
  claude
  committed
  2073451
- fix(evals): simulated task results keep a finished status and follow T3's target choice
  
  only21mil
  and
  claude
  committed
  8c7a8af