All pull requests

Pull requests

• feat(harness): record why a rebuild drops a scenario for a removed capability and restore it under its key when the capability returns (TH-8754)

  #325·KarthikAvinashFI opened Nov 15, 2026Member·

  ·
  

• fix(harness): reseal the scenarios that load and name the ones that could not be applied (TH-8929)

  #324·KarthikAvinashFI opened Nov 15, 2026Member·

  ·
  

• test(harness): green the harness suite and run it in CI

  #323·KarthikAvinashFI opened Nov 14, 2026Member·

  ·
  

• feat(harness): seal a stopped call with its partial evidence, flag worker channel bypass and record shared records (TH-10189)

  #322·KarthikAvinashFI opened Nov 14, 2026Member·

  ·
  

• feat(harness): negative controls, leaked-answer check, grounding and suite proving (TH-8918)

  #321·KarthikAvinashFI opened Nov 14, 2026Member·

  ·
  

• fix(harness): save whether each proof's solution ran and report proved counts (TH-8752)

  #320·KarthikAvinashFI opened Nov 14, 2026Member·

  ·
  

• fix(harness): run tool handlers as the agent's user and give each world its own database role

  #319·NVJKKartik opened Nov 13, 2026Member·12 tasks done·

  ·

• fix(harness): name a served coordinator's own calls by its run, as its jobs' hand-offs do

  #318·KarthikAvinashFI opened Nov 13, 2026Member·

  ·
  

• feat(harness): report every builder chat turn's usage with its ending

  #317·NVJKKartik opened Nov 12, 2026Member·

  ·
  3

• fix(harness): run each customer process as its own OS user when the sandbox grants it

  #316·NVJKKartik opened Nov 12, 2026Member·12 of 15 tasks·

  ·
  9

• fix(harness): name why a candidate failed certification with a closed cause

  #315·NVJKKartik opened Nov 12, 2026Member·7 of 11 tasks·

  ·
  9

• feat(harness): measure an untraced agent's model calls in the sandbox

  #314·NVJKKartik opened Nov 12, 2026Member·17 of 22 tasks·

  ·
  17

• fix(harness): pass only the values a hosted run uses through hosted entry points

  #313·nik13 opened Nov 12, 2026Collaborator·5 of 10 tasks·

  ·

• [TH-9550] fix(harness): load a repository's .env into hosted authoring only through the allowlist

  #312·khushalsonawat opened Nov 12, 2026Member·1 task done·

  ·
  1

  

• test(harness): a CrewAI reference agent for the hosted path (TH-8811)

  #311·khushalsonawat opened Nov 12, 2026Member·7 of 9 tasks·

  ·
  6

• feat(harness): run the contract save's file checks in a child with a deadline (TH-9445)

  #310·khushalsonawat opened Nov 12, 2026Member·

  ·

• test(harness): the worker inherits the runner's Cartesia key, and CI runs the worker-secrets tests

  #309·khushalsonawat opened Nov 12, 2026Member·2 tasks done·

  ·
  2

  

• fix(harness): catch secrets by shape in the kit's one cleaner (TH-9500)

  #308·khushalsonawat opened Nov 12, 2026Member·2 tasks done·

  ·

• [TH-9550] fix(harness): keep job-chosen names out of the sandbox controller's environment

  #307·khushalsonawat opened Nov 12, 2026Member·1 task done·

  ·
  1

  

• test(harness): a Deep Agents reference agent for the hosted path (TH-8810)

  #306·khushalsonawat opened Nov 11, 2026Member·7 of 9 tasks·

  ·
  5

• test(harness): a Google ADK reference agent for the hosted path (TH-8809)

  #305·khushalsonawat opened Nov 11, 2026Member·7 of 9 tasks·

  ·
  7

• test(harness): a Pydantic AI reference agent for the hosted path (TH-8808)

  #304·khushalsonawat opened Nov 11, 2026Member·7 of 9 tasks·

  ·
  6

• fix(harness): let an OpenAI Agents SDK agent run on the model gateway (TH-8800)

  #303·khushalsonawat opened Nov 11, 2026Member·6 of 9 tasks·

  ·
  6

• fix(simulate): move Retell web calls to the v3 gateway (TH-8763, stacked from #295)

  #302·khushalsonawat opened Nov 11, 2026Member·1 task done·

  ·
  

• test(harness): an OpenAI Agents SDK reference agent for the hosted path (TH-8800)

  #301·khushalsonawat opened Nov 11, 2026Member·7 of 9 tasks·

  ·
  7