Skip to content

Vally 0.16.0

Released on 2026-09-08.

Breaking

  • Grader name values must now be unique kebab-case slugs of at most 60 characters. (#1107)
  • Public API (@microsoft/vally): ToolCallEvent.data.arguments can now be any JSON root value rather than only an object. (#1009)
  • vally-server schema advances to v8 to distinguish grader execution errors from failing verdicts. Delete existing analytics databases and re-ingest their source JSONL. (#1072)

Features

  • LLM grading fails explicitly when the selected evidence resolves to nothing instead of silently treating the result as empty. (#1083)
  • Linting now catches invalid directory-discovery config and stops silently skipping evals in broken directory setups. (#1135, #1137)

Fixed

  • Copilot session logging is isolated per filesystem so agent runs no longer leak or overwrite one another’s temp state. (#1109)
  • Export logs preserve the output directory in generated paths, and Windows subprocesses no longer surface console windows. (#1126, #1117)