Turn on search by meaning from settings or the command palette. Keyword search takes over when meaning-based search is unavailable.
Trace a commit or file line back to the session and prompt that produced it.
Usage and activity pages load faster, with remote PostgreSQL page latency cut by up to 72%.
Import xAI billing exports and Open Interpreter sessions. Claude Code hook output appears in collapsible cards, and peer messages stay in transcripts.
Fork messages remain visible when parent sessions are missing, archive rebuilds keep moving, and one failed dashboard panel no longer breaks the others.
Find pull requests and issues by meaning from the command palette, with optional embeddings for semantic search.
Filter pull requests, issues, merge queues, worklists, and Activity by labels. Worklists also support involvement and readiness filters.
Merge after CI now uses GitHub auto-merge, so queued merges survive Forge restarts. You can also remove items from the queue or turn auto-merge off.
Large repositories load faster across summaries, pull request lists, Activity, mention suggestions, and diffs. A repository summary with 1.8 million cached items now loads in about a second.
Forge now shows hidden merge blockers even when every listed check is green.
Manage recurring work from the CLI or terminal UI, including starting, changing, and stopping a recurrence.
Turn on search by meaning in Settings, with one-command setup for local embeddings and better matching across comments.
Preview and move issues between projects from the terminal UI, or delete and restore them without losing history.
Create and edit cron jobs and workflows from the Jobs tab, and see the next three scheduled run times.
Large issue lists and graphs now render without freezing the browser, while issue reads, duplicate checks, and federation syncs do less unnecessary work.
You can now read and mark PDFs in one browser workspace, mask selected text, and save reviewed copies with Bates numbers.
Docbank now accepts linearized and Office-saved PDFs, turns review pages correctly, and redacts cropped PDFs with fractional geometry.
You can now enable local text search with one command, extract PDF text locally, and keep existing Nomic vectors during later setup runs.
PostgreSQL storage now supports photo imports, albums, review, face and people curation, metadata editing, previews, recordings, and email attachments.
Agents can now cite evidence and coverage in answers, flag conflicting figures across sources, and resume streamed vault changes from a cursor.