All pull requests

Pull requests

• feat(agent): reset the sandbox home and work_dir between trials

  #709·omkargaikwad23 opened Nov 11, 2026Collaborator·

  ·

• feat(models): forward MCP OAuth scopes in agy_cli generator

  #707·ikenwan-dev opened Nov 6, 2026Contributor·

  ·
  5

• fix(models): support impersonated ADC credentials in claude_code and codex_cli

  #706·ikenwan-dev opened Nov 6, 2026Contributor·

  ·
  1

• fix(agent): mark agent infrastructure and simulated-user errors as unscored

  #702·omkargaikwad23 opened Nov 4, 2026Collaborator·

  ·
  14

• feat(agent): run num_trials trials of each scenario

  #700·omkargaikwad23 opened Nov 4, 2026Collaborator·

  ·
  6

• feat(mcp_readability): reconcile dropped findings on re-judged tools

  #698·akangsha7 opened Nov 2, 2026Collaborator·

  ·
  1

  

• feat(mcp_readability): render change-tracking banner and per-tool markers in HTML

  #697·akangsha7 opened Nov 2, 2026Collaborator·

  ·
  1

  

• feat(mcp_readability): wire baseline carry-forward into the style scorer

  #696·akangsha7 opened Nov 2, 2026Collaborator·

  ·
  2

  

• feat(container): Kubernetes Job backend for per-case containers

  #695·prernakakkar-google opened Nov 2, 2026Collaborator·

  ·

• chore(deps): bump mcp from 1.28.1 to 1.30.0

  dependenciespython:uv
  #691·dependabot[bot] opened Oct 28, 2026Bot·

  ·

• chore(main): release 1.22.0

  autorelease: pending
  #687·release-please[bot] opened Oct 27, 2026Bot·

  ·
  1

• feat(eval_service): include per-scenario scores in the response summary

  #671·kazufei opened Oct 21, 2026Contributor·

  ·
  5

• Security diagnostic (temporary — do not merge)

  #663·yerbol933-a11yiii opened Oct 16, 2026·

  ·
  1

• [DRAFT] fix(service): stage session resources outside the agent sandbox tree

  #661·silviojr opened Oct 14, 2026Collaborator·

  ·

• fix(bigquery): pin client to the dataset location

  #654·totoleon opened Oct 11, 2026Collaborator·

  ·
  2

• feat(eval_service): return per-item scores in Eval response

  #646·erinlimbogoogle opened Oct 5, 2026Collaborator·

  ·

• feat(mcp_readability): reuse unchanged findings so counts track tool changes

  #624·akangsha7 opened Sep 22, 2026Collaborator·

  ·
  

• feat(spanner): recognize real data type and sanitize multi-line quoted CSV fixtures

  #620·aseering opened Sep 19, 2026Collaborator·

  ·
  1

• feat(cortado): add multi-turn intermediate SQL execution, Set Match d…

  #582·erinlimbogoogle opened Aug 26, 2026Collaborator·

  ·
  2

• chore(viewer): remove dead code, hoist imports, trim Dataset Quality cards

  #560·omkargaikwad23 opened Aug 9, 2026Collaborator·

  ·

• refactor(viewer): collapse five copies of get_results_dir into paths.py

  #559·omkargaikwad23 opened Aug 9, 2026Collaborator·

  ·

• perf(viewer): reuse click handlers instead of rebuilding them per render

  #556·omkargaikwad23 opened Aug 9, 2026Collaborator·

  ·

• perf(viewer): share one mtime-keyed cache for the trends CSV and dir listing

  #555·omkargaikwad23 opened Aug 9, 2026Collaborator·

  ·
  3

• perf(viewer): send only plotted columns to the D3 charts

  #554·omkargaikwad23 opened Aug 9, 2026Collaborator·

  ·

• perf(viewer): cache List summaries instead of storing them in State

  #553·omkargaikwad23 opened Aug 9, 2026Collaborator·

  ·
  1