NovaCode 1.2.3

NovaCode 1.2.3

Visible, selective memory recall

• Show “Recalling memory” inline when relevant topic excerpts are automatically
  supplied to the model, with the source path when backed by a file.
• Show “Reading memory” after a successful explicit memory-page read, including
  reads made through Code Mode. Failed reads and search results do not claim a
  page was read.
• Keep standing preferences and the compact index available automatically.
  Search summaries and read relevant full pages when more detail is needed,
  rather than loading the whole memory wiki.
• Suppress duplicate notices for the same source and action within a turn.
  Notice events carry source metadata, not memory bodies, and do not add
  messages to model history. Existing permissions and retrieval limits apply.

Memory sources

• Return the actual project-scoped topic path from memory search and automatic
  recall. General and project topics with the same name remain distinct.
• Identify generated knowledge without inventing a file path to read.

Verification

• Add model-boundary, normal-tool, Code Mode, shared-stream and TUI regression
  coverage for memory notices.
• Fix workspace-mock cleanup in resume tests so later async-agent tests use
  their own workspace correctly.

NovaCode 1.2.2

NovaCode 1.2.2

Session resume

• Browse saved conversations across projects with /resume and nova --resume.
• Choose the saved workspace or current workspace when resuming another
  project's session, with optional “Always use” preferences. Reset the choice
  with nova config set resume_directory ask.
• Open the saved workspace in an isolated session tab while preserving the
  current chat. Choosing the current workspace restores the conversation in place.
• Preserve workspace trust and tool approvals, context-mode choices and saved
  session identity. Switch to an already-open session instead of loading it twice.
• Keep the existing 200-session browser limit and exclude cleared sessions.

Fixes

• Fix the Textual widget error when displaying a compacted conversation summary.

Packaging

• Show the release summary and changelog link on PyPI, and attach this changelog
  to the GitHub release.