github matthiasn/lotti 1.1.70+4458

pre-release4 hours ago

What's Changed

  • feat(search): index every journal write; ranked, filtered full-text matching that never throws by @matthiasn in #4872
  • feat(search): jump to tasks, entries, projects, events and people by title from the command palette by @matthiasn in #4873
  • fix(goals): the goal agent answers what the user said in their check-ins by @matthiasn in #4874
  • feat(search): task search lists the best match first, with where each task matched by @matthiasn in #4875
  • test(goals): journaling evals for the user-voice contract and the tail budget by @matthiasn in #4876
  • docs(search): choose the on-device embedding model on Lotti's own retrieval task (ADR 0144) by @matthiasn in #4877
  • test(goals): live journaling evals on glm-5.3; the journal tail stays 1,200 by @matthiasn in #4878
  • fix(search): a task keeps one agent report in the vector index, however many embeds were skipped by @matthiasn in #4879
  • chore(release): 1.1.70 by @matthiasn in #4880

Full Changelog: 1.1.69+4457...1.1.70+4458

Don't miss a new lotti release

NewReleases is sending notifications on new releases.