Skip to content

v0.3.0

Choose a tag to compare

View all tags
@para1992 para1992 released this 16 Aug 16:27
· 21 commits to main since this release
v0.3.0
afbbfa5

What's new

  • Automatic judge (judge()) — one structured LLM invocation selects the best candidate and the run completes without a human review barrier. Replace manualJudge() with judge() and outputGlue('draft') keeps the selected result.
  • await_human step — a schema-validated human approval barrier for workflows that need an explicit sign-off.
  • rick.humanizer recipe — language-neutral text humanization with pattern, taste, and fidelity audits plus factual grounding against the source. Optional voice-sample calibration via use_voice_sample.
  • Prompt-limit handling — prompt overflows now fail the step cleanly through PromptLimitExceededException (a StepFailureBase).
  • Migration repair — invocation-diagnostic backfill rows skipped by the 0.1.0/0.2.0 migration are repaired by the new 000006 migration; existing installs stay untouched.

Docs

  • README now covers judge() and the humanizer recipe under "Typical tasks".
  • CHANGELOG.md updated with the [0.3.0] section.

Full changelog: https://github.com/para1992/laravel-rick/blob/main/CHANGELOG.md

Assets 2
Loading