Skip to content

v1.47.0

Choose a tag to compare

@futureagi-release-bot futureagi-release-bot released this 05 Oct 11:07
cbeaa0e

1.47.0 (2026-10-05)

Features

  • add call and voice metric tiles to run analytics (ee4cc51)
  • simulate: collapsible, resizable builder chat on the environment page (0605508)
  • simulate: collapsible, resizable builder chat on the environment page (#3260) (7e29023)

Bug Fixes

  • align simulation call outcomes and run details (4e178c3)
  • evals: count wrong-language, wrong-product and unanswered replies as unhandled queries (2507764)
  • evals: generalise customer-agent criteria and give harness evals the full prompt and correct speaker labels (8edb063)
  • evals: generalize and shorten conversation criteria (e113b27)
  • evals: keep the added-eval action button still while it loads (ffb29d4)
  • evals: keep the added-eval action button still while it loads (524472b)
  • evals: tighten conversation criteria from voice-call audit (8aa8bd3)
  • evals: tighten generic conversation criteria (1a641c8)
  • simulate: builder chat collapse/resize and run calls toolbar wrap (#3260, #3299) to main (0f59200)
  • simulate: drop the collapsed chat's dot once the chat can't take an answer (dbedcd6)
  • simulate: hide only reliability behind the detailed analytics toggle (aab6ba4)
  • simulate: hide only the reliability section behind the detailed analytics toggle (d39d4ae)
  • simulate: keep run analytics up when the response has no reliability block (116bef8)
  • simulate: keep the chosen chat width through window resizes and announce the real maximum (aa6ff43)
  • simulate: keep the chosen chat width when the divider is used in a narrow window; count the divider in the max (6bae721)
  • simulate: keep the collapsed chat's dot on while the builder waits for an answer (8e2f8d3)
  • simulate: keep the run calls toolbar tidy at narrow widths (24fc3bd)
  • simulate: keep the run calls toolbar tidy at narrow widths (62d8d42)
  • simulate: keep the run calls toolbar tidy at narrow widths (#3299) (7e30a4d)
  • simulate: leave the run calls group header's partial counts empty (a555999)
  • simulate: leave the run calls group header's partial counts empty (aebd97c)
  • simulate: lock and re-read each call before the backfill rewrites its metadata (8557cb0)
  • simulate: preserve full Others agent prompt (71f2c07)
  • simulate: restore the persona count and partial status dash in the group header (a24c0e0)
  • simulate: restore the persona count in the run calls group header (1366080)
  • temporal: leave register_temporal_schedules without interpreter teardown (6e2fdd4)
  • temporal: leave register_temporal_schedules without interpreter teardown (066d8b3)
  • tracer: log grouping control conflict reasons (0f77b87)
  • tracer: log grouping control conflict reasons (4c2914c)
  • tracer: stop the PeerDB Temporal setup job crashing as it exits (01d0829)
  • tracer: stop the PeerDB Temporal setup job crashing as it exits (05e2707)

Performance Improvements

  • simulate: format a tool row's output once per result, not on every render (e0749dd)

Reverts

  • evals: keep the termination handling criteria as on dev (a1dda02)