Skip to content
Vally

Vally 0.1.0

Released on 2026-04-08.

Changes

  • Adding Microsoft SECURITY.MD (#2)
  • Add README (#4)
  • Add MIT License to the project (#5)
  • Adding Microsoft SECURITY.MD (#8)
  • Add MIT License to the project (#10)
  • Fix eval --workers help text, consolidate int flag parsing (#11)
  • Updating, trying to make it so we can CI as a check. (#12)
  • grade accepts JSONL input from eval (#14)
  • feat(core): scaffolding module with typed eval builder (#17)
  • feat(core): Interactive skill wizard module (E-4) (#19)
  • feat(cli): init and new commands (E-5) (#20)
  • refactor(core): unified discovery and exports (E-6) (#21)
  • feat(core): project configuration module with JSON schema (#26)
  • Ensure eval JSONL output includes errors from failed stimuli (#28)
  • Add license headers (#29)
  • Enhancing tool-grader to allow additional parameters, like matching against the command or path (#31)
  • Add eval spec validator (pre-flight checks) (#34)
  • fix: validator review feedback and lint tests in CI (#35)
  • Add LLM-as-a-judge glossary and AGENTS.md (#36)
  • grade prints human-readable summary to stderr in JSONL mode (#40)
  • Initial commit of docs site. (#42)
  • Allow for regex flags (?ims) in our pattern based fields (#43)
  • feat: LLM-as-judge prompt grader (#44)
  • Fixing a bug where the executor wasn’t actually being controlled by configuration (#48)
  • feat: pairwise comparative LLM-as-judge grader + compare command (#49)
  • eval output reporter refactoring (#50)
  • Renaming things to be vally, for now so we can publish a package. (#52)
  • Executor needs to be stopped, and type checking for tests (#54)
  • Fix console indentation of multi-line evidence (#59)
  • Named Environments (#60)
  • Eval Tags & Suites (#61)
  • Add ADO Publishing Pipeline (#62)
  • Make it possible to start off your tmpdir with a git worktree (#64)
  • Prevent init and new tests writing to stdout (#65)
  • Add agent instruction about license headers (#66)
  • grade requires an eval spec (#67)
  • build(deps): bump vite from 7.3.1 to 8.0.5 (#70)
  • build(deps): bump vite from 7.3.1 to 7.3.2 in /docs-site (#72)
  • build(deps): bump defu from 6.1.4 to 6.1.6 in /docs-site (#74)
  • Trim architecture.md now that we have a docs site. (#75)
  • feat: completed/exit-success grader, negation variants, validator fix (#76)
  • feat: add file-matches grader (regex matching on file content) (#77)
  • Format eval output consistently (#78)
  • Copy env.skills into workspace (#80)
  • vally CLI (#81)
  • There were more skill-validator references in…everywhere. (#83)
  • Grader names are kebab-case (#84)
  • One more docs change, some skill-validator references were intentional. (#85)