v0.1.44-beta.14

v0.1.44-beta.14

Added

• Run automatic memory learning through a persisted, bounded background worker so
  completed chat turns no longer wait for the memory provider; expose extraction
  outcomes, Dream maintenance, write policy, and Markdown sharing controls.
• Add recoverable memory maintenance and revision-aware Markdown import/export
  with explicit opt-in scopes, validation, and visible conflicts.

Changed

• Keep GPT-6 Astra compatible with the current Codex Responses Lite protocol and
  surface normalized multi-provider cache-hit usage to prompt status integrations.
• Make the Minimal prompt preset truly compact and retain the final Arcade/Snake
  board after compaction without leaving timers or input focus active.

Fixed

• Preserve queued prompt delivery and terminal cancellation while compaction,
  background extraction, reconnect recovery, or long-running shell work settles.
• Keep provider startup usable when metadata loading fails or a valid catalog is
  empty, and report configuration errors without trapping Home in startup state.
• Carry forward database compatibility checks and Blacksmith-backed security and
  startup regression workflows from the latest public branch.

Tests

• Add focused regressions for memory workers, Dream and Markdown sharing, provider
  startup, Astra OAuth compatibility, cache reporting, compact prompt layout,
  final compaction rendering, queue delivery, and repeated cancellation.

Experimental prerelease from d080b73. Concurrency remains opt-in.

Verify

Built from d080b73fe2b7f852c3389b68e07858e6088d6922.

gh release download v0.1.44-beta.14 --repo MendCode/MendCode
shasum -a 256 -c SHA256SUMS
for file in mendcode-*.zip mendcode-*.tar.gz SHA256SUMS RELEASE-MANIFEST.txt release-index.json mendcode.spdx.json; do
  gh attestation verify --repo MendCode/MendCode "$file"
done