All pull requests

Pull requests

• refactor(www): generate API types from the OpenAPI spec

  #1532·theSatvik opened Aug 29, 2026Contributor·5 tasks done·

  ·
  1

  15

• feat: run large eval sets across multiple runner pods

  #1518·spruceb opened Aug 28, 2026Contributor·5 tasks done·

  ·
  51

• feat(api): refuse eval-set submissions from a user whose runs cannot start

  #1516·rasmusfaber opened Aug 28, 2026Contributor·5 tasks done·

  ·
  5

• feat(ops): move a model between access groups and reconcile the caches

  #1508·PaarthShah opened Aug 27, 2026Contributor·4 tasks done·

  ·
  9

• fix(datadog): page on event-loop saturation only when model calls are timing out

  #1507·PaarthShah opened Aug 27, 2026Contributor·

  ·
  3

• feat(sample-edit): reject a closed-vocabulary score edit at the API boundary

  #1505·sjawhar opened Aug 27, 2026Contributor·5 tasks done·

  ·
  16

• fix(infra): scope private hosted zone to hawk.<domain> instead of the base domain

  #1490·dmitrii opened Aug 27, 2026Contributor·4 of 5 tasks·

  ·
  1

  6

• feat: add conditional scan workflows

  #1484·love-lena opened Aug 26, 2026Contributor·

  ·

• chore(lint): standardize ruff checks and enable FURB118

  #1473·PaarthShah opened Aug 26, 2026Contributor·5 tasks done·

  ·
  7

• feat(cli): use legible model IDs in downloaded artifact folder names

  #1469·spruceb opened Aug 26, 2026Contributor·

  ·
  1

  1

• fix: hint aggregator model names when eval-set validation fails

  #1457·apurv-1 opened Aug 25, 2026Contributor·5 tasks done·

  ·
  1

  4

• fix(runner): make hawk stop end evals wedged in sandbox setup

  #1437·rasmusfaber opened Aug 24, 2026Contributor·5 tasks done·

  ·
  8

• feat(runner): default ephemeral-storage limits for sandbox pods

  #1433·madasigon opened Aug 24, 2026Contributor·

  ·
  1

  8

• fix(importer): count samples when an eval log has no results block

  #1426·sjawhar opened Aug 23, 2026Contributor·5 tasks done·

  ·
  8

• feat(cli): confirm what hawk stop and hawk delete did

  #1424·madasigon opened Aug 22, 2026Contributor·5 tasks done·

  ·
  1

  11

• fix(monitoring): bound the provider's Kubernetes API calls [HWK-200]

  #1418·dmitrii opened Aug 21, 2026Contributor·3 tasks done·

  ·
  5

• fix(middleman): split the OIDC audience config middleman already accepts as a list

  #1413·sjawhar opened Aug 21, 2026Contributor·5 tasks done·

  ·
  11

• fix(middleman): stop paging Sentry for handled upstream openai 4xx [SEN-235]

  #1406·metr-background-agents[bot] opened Aug 20, 2026Bot·5 tasks done·

  ·
  3

  

• fix(middleman): reject deprecated Gemini labs with a 400, not an unhandled 500 [SEN-236]

  #1405·metr-background-agents[bot] opened Aug 20, 2026Bot·5 tasks done·

  ·
  1

• fix(middleman): stop 500ing on models whose lab has no dispatch class [SEN-237]

  #1404·metr-background-agents[bot] opened Aug 20, 2026Bot·5 tasks done·

  ·
  3

• fix(middleman): keep secret model codenames out of the traffic log

  #1393·rasmusfaber opened Aug 20, 2026Contributor·5 tasks done·

  ·
  2

• feat(timeline): chart model and tool call duration over time

  #1385·pipmc opened Aug 19, 2026Contributor·5 tasks done·

  ·

• feat(infra): alarm on middleman traffic-log data loss

  #1353·rasmusfaber opened Aug 18, 2026Contributor·6 tasks done·

  ·
  6

• perf(viewer): debounce jobs search requests

  #1338·rasmusfaber opened Aug 17, 2026Contributor·4 of 5 tasks·

  ·
  2

• ci: fail fast on colliding Alembic migration heads

  #1334·dmitrii opened Aug 17, 2026Contributor·5 tasks done·

  ·
  18