Skip to content

v1.5.0 - faster recall, forget controls, and Antigravity support

Latest

Choose a tag to compare

@stoicsoft stoicsoft released this 05 Jul 05:47

1AIVault now recalls the right memories faster, captures live AI conversations with less friction, and gives you a reversible way to keep unwanted context out of normal vault results.

New Features

forget_entry
forget_topic

Forget and Remember

  • Forget individual memories — Hide an entry from search, topics, imports, MCP reads, and injected memory without deleting it
    • Restore forgotten memories from Settings whenever you need them back
    • Future imports of the same source conversation stay hidden until you remember it
  • Forget entire topics — Hide a topic and its assigned memories in one action when a project, thread, or old context should stop reaching your AI tools
    • Remembering a topic brings its timeline back while entries you forgot individually stay hidden
  • Let AI clients forget on request — Claude Desktop, Claude Code, Cursor, Cline, and other connected clients can forget or remember entries and topics through the vault when you explicitly ask

Antigravity CLI Support

  • Import Antigravity conversations — Bring Antigravity CLI sessions into the vault alongside Claude Code, Codex, Gemini, Cursor, Cline, and OpenCode
  • Chat with Antigravity — Route a Unified Chat turn to Antigravity, pick from detected Antigravity models, and see Antigravity activity with the right source label
  • Classify Antigravity memory — Use Antigravity in Classify Now and see Antigravity entries across topics, activity, memory reads, and related views

Faster Live Capture

  • Capture Claude Code and Codex sooner — When live sync is enabled, new Claude Code and Codex sessions can land in the vault shortly after they finish
    • If 1AIVault is closed, new session notices are saved and processed on the next launch
    • The existing background scan still backfills anything missed

Improvements

  • Much faster recall on large vaults — Search results now stay compact, show the relevant part of each memory, and avoid sending full saved conversations unless an AI tool asks for them
  • Better memory ranking — Recent and frequently used memories get stronger placement, pinned memories remain first, and tag matches are now searchable
  • Smarter related memory — Related-memory lookups and project-scoped injected context can include nearby memories that use different wording when semantic recall is enabled
  • Smoother live imports — Claude Code, Codex, Cursor, OpenCode, and Antigravity do less repeat work while watching active conversation folders, so busy setups feel more responsive
  • Claude Code session continuity — Resumed and compacted Claude Code sessions stay connected to their original thread so one logical conversation is easier to recall
  • Stronger auto-injected briefs — New session context now includes recent work and classification gaps, skips forgotten memory, and better tells AI tools when to search, save, import, or classify
  • Clearer Connection guidance — Auto-Inject now shows context freshness, classification gaps, and import coverage so you can refresh what your AI tools know before starting a session
  • Faster app navigation — Heavy screens such as Chat, Skills, Collect, Classify, Connection, Settings, command palette, and update dialog load only when opened
  • Better Classify workspace — Resize the entry list, open entry details fullscreen, and use cleaner AI tool logos while sorting through large vaults

Bug Fixes

  • Fixed tag searches missing tagged-only memories — entries now appear when the matching word is in a tag, even if it is not in the memory body
  • Fixed search previews starting at the beginning of long conversations — previews now jump to the matching part of the memory so results are easier to choose
  • Fixed AI searches causing needless instruction file rewrites — searching from connected clients no longer refreshes auto-injected files when no memory content changed
  • Fixed large transcript imports slowing down linking — long raw conversations skip low-value per-entry linking while useful tag-based connections still build in the background
  • Fixed deleted local transcripts disturbing saved vault memory — when an AI tool cleans up its own conversation files, existing vault entries stay available

Download: https://1aivault.com