All pull requests

Pull requests

• fix(optimize): cap-evolve integration fixes from #255 review

  #288·OsherElhadad opened Oct 28, 2026·6 tasks done·

  ·
  15

• feat(runner): stream fullsend's console live under the progress prefix

  risk/moderate
  #286·astefanutti opened Oct 28, 2026Collaborator·

  ·
  11

• fix(runner): follow a child harness's base: chain for the image pin

  ready-for-mergeready-for-reviewrisk/moderate
  #284·fullsend-ai-coder[bot] opened Oct 28, 2026Bot·4 tasks done·

  ·
  1

  3

  

• docs(spec): 016 — one field for "this case produced no measurement"

  kind/documentationrisk/moderatespex/spec
  #280·astefanutti opened Oct 27, 2026Collaborator·

  ·
  33

• [WIP]feat(agent-eval-optimize): cap-evolve integration

  #255·AIMikav opened Oct 15, 2026·

  ·
  24

• feat: measurement-validity program — chance-corrected reliability across all three eval layers (arXiv 2608.00794)

  #200·astefanutti opened Aug 21, 2026Collaborator·

  ·
  4

• ci: add harness-eval static analysis for agent configurations

  #189·Benkapner opened Aug 17, 2026Contributor·1 of 2 tasks·

  ·
  5

• feat(score): reasoning-effort knob for LLM judges

  #188·astefanutti opened Aug 14, 2026Collaborator·

  ·
  1

• refactor(report): replace Markdown renderer with Mistune

  #173·neallin917 opened Aug 4, 2026Contributor·

  ·
  1

  8

• fix(harbor): root-cause + fix recurring verifier no-reward flake (hung exec establishment)

  #140·astefanutti opened Jun 26, 2026Collaborator·

  ·
  3

• feat(eval-replay): add skill for outcome-anchored evaluation from historical PRs

  #134·smith-xyz opened Jun 22, 2026·3 tasks done·

  ·
  29

• Add download and copy-path toolbar to HTML eval report

  #123·jaquevan opened Jun 12, 2026·7 tasks·

  ·
  1

• feat: add agent-eval run CLI for deterministic eval execution

  #109·astefanutti opened Jun 4, 2026Collaborator·3 of 5 tasks·

  ·
  1

• fix(trace_builder): populate mlflow.llm.cost and gateway.endpointId for cost charts

  #101·abhijeet-dhumal opened Jun 2, 2026Contributor·

  ·
  1

  7

• fix(log_results): handle AsyncExportQueue rename in MLflow >=3.13

  #100·abhijeet-dhumal opened Jun 2, 2026Contributor·

  ·
  1

  6

• fix(score): use final result text as conversation content for judges

  #98·abhijeet-dhumal opened Jun 2, 2026Contributor·

  ·
  1

  2

• feat: OTel trace capture and OpenCode runner

  #90·astefanutti opened Jun 1, 2026Collaborator·3 of 5 tasks·

  ·
  11

• feat(eval-optimize): add optimization controls for robust skill improvement

  #73·astefanutti opened May 27, 2026Collaborator·5 tasks·

  ·
  13

  

• Add --bare mode for reproducible headless execution

  #67·astefanutti opened May 18, 2026Collaborator·1 of 2 tasks·

  ·
  7