All pull requests

Pull requests

• agent: add move status reporting (clipping/stall/ok) [#408]

  #553·Atif0110 opened Oct 5, 2026·8 tasks·

  ·
  1

• feat(frames): persist metric depth next to stored RGB frames

  #552·VARUN3WARE opened Oct 4, 2026·8 tasks done·

  ·
  2

• docs: list community plugins, starting with inspect-robots-wandb

  #551·Galabavamsi opened Oct 3, 2026Contributor·

  ·
  3

• feat(eval): optional on_eval_error sink hook for aborted runs

  #550·Galabavamsi opened Oct 3, 2026Contributor·

  ·
  5

• feat(agent): report every move's outcome with per-axis shortfall and approver cause

  #549·VihanAggarwal opened Oct 3, 2026·8 tasks done·

  ·
  1

  1

• feat(evidence): uncertainty and scene-paired comparison for saved runs

  #548·VihanAggarwal opened Oct 3, 2026·8 tasks done·

  ·
  3

• fix(log): keep NumPy scalars in scene metadata

  #547·VanshBhardwaj24 opened Oct 3, 2026Contributor·6 of 8 tasks·

  ·
  2

• fix(setup): keep active config and clean temp files on replacement failure (#488)

  #546·shobhitagnihotri69 opened Oct 3, 2026Contributor·

  ·
  1

  3

• fix(eval): remap observation keys to policy and reject collisions (#494)

  #545·shobhitagnihotri69 opened Oct 3, 2026Contributor·

  ·
  1

  3

• fix(eval): record effective trial seeds and scene init_seed in saved …

  #544·shobhitagnihotri69 opened Oct 3, 2026Contributor·

  ·
  1

  4

• fix(conformance): filter out OptionSlot entries with non-boolean defaults in option_slots

  #534·Adityakk9031 opened Oct 3, 2026Contributor·

  ·
  1

• Resume evaluation sets from explicit checkpoints

  #514·Galabavamsi opened Oct 1, 2026Contributor·8 tasks done·

  ·
  1

  8

• feat(eval): add parallel simulation eval sets

  #510·agocharbhatia opened Sep 29, 2026·

  ·
  1

  4

• fix(summarize): safely fall back when transcripts are missing in build_digest

  #500·Adityakk9031 opened Sep 26, 2026Contributor·

  ·
  4

• fix(chatwire): normalize response read failures

  #487·DivyamTalwar opened Sep 25, 2026Contributor·8 tasks done·

  ·
  1

  1

• fix(agent): remove deepcopy in _sanitize to prevent massive memory footprint

  #485·Adityakk9031 opened Sep 25, 2026Contributor·

  ·
  4

• fix(eval): catch unhandled rollout exceptions to prevent log loss

  #484·Adityakk9031 opened Sep 25, 2026Contributor·

  ·
  8

• ▚▚ fix(task): resolve max_seconds without binary float rounding

  #483·breken-ai opened Sep 25, 2026Contributor·

  ·
  3

• fix(chatwire): reject explicitly truncated completions

  #482·DivyamTalwar opened Sep 25, 2026Contributor·8 tasks done·

  ·
  1

  1

• Review trial: restore Sravanthi's Jev policy plugin (#453)

  #456·jeqcho opened Sep 20, 2026Collaborator·

  ·
  25

• Add advisory PR reviews with Codex on Cloudflare

  #455·jeqcho opened Sep 20, 2026Collaborator·

  ·
  16

• fix(eval): release string-resolved and CLI-invoked policies in finally block

  #446·Adityakk9031 opened Sep 13, 2026Contributor·

  ·
  13

• fix(console): prevent Windows stdin select crash and add msvcrt fallback

  #445·Adityakk9031 opened Sep 13, 2026Contributor·

  ·
  13

• core: collision-proof run identity and final logs

  #435·ethanvillalovoz opened Sep 4, 2026·8 tasks done·

  ·
  1

  2

• fix(eval): preserve stop signal across cleanup failure

  #429·Vedangalle opened Aug 31, 2026Contributor·9 tasks done·

  ·
  5