Skip to content

v2.0.0

Choose a tag to compare

View all tags
@openadapt-release openadapt-release released this 15 Oct 09:01
· 24 commits to main since this release
Immutable release. Only release title and notes can be modified.
v2.0.0
04fe6fe
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

v2.0.0 (2026-10-15)

This release is published under the MIT License.

⚠ Breaking change: every saving run is gated by default

  • openadapt-flow run now uses the each_item approval mode unless the run names a mode (#497). Before 2.0.0 only the Standard and Regulated profiles were gated by default; a Demo or profile-less run saved without asking.
  • What you will notice: a run that reaches a consequential save now pauses until a person approves that entry (openadapt-flow item-approval, Desktop, or an attended agent), instead of saving straight away.
  • To keep the old behavior for a run, opt out explicitly: openadapt-flow run BUNDLE --approval-mode not_gated, or set runtime.approval_mode: not_gated in the deployment config (--config). The run report records that the run was not gated. --approval-mode problems_only sends only problems to a person.
  • A resumed run keeps its own recorded mode.

Features

  • Attach Cloud's signed Execute receipt (the Seal) to a VERIFIED run report: run --cloud-execution-id and attach-seal (#506).
  • Send repair candidates to Cloud review and take decisions back (#502).
  • DOM page-structure identity for form fields (#503).
  • A click that only opens a confirmation dialog is not the save (#499).
  • Identify web apps by address and signed-in session (#498).
  • Run the UI read-back as a pilot under per-entry approval only (#500).
  • A person approves each save; readiness drills run from Desktop (#486).
  • Bind an unlabeled Save to the form it submits (#496).
  • Let a private pack check the saved record through deployment config (#495).
  • Check a save through the app's record page in a separate session (#492).
  • One-command demo page and truthful first-run wording (#483).

Performance

  • About 29% faster OpenEMR referral replay without weakening checks (#501).

Bug Fixes

  • Carry the recorded fix target and place the fix after a step that acted (#494).
  • Size the attached test browser's real window (#493).
  • Fail closed on uncertain run outcomes and redact emitted skills (#484).
  • Treat an icon-only Save as a save; start runs in the signed-in browser (#487).
  • Stop false right-record halts on changed names; repair the OpenEMR harness (#485).
Assets 5
Loading