Skip to content

Storyworks Workspace v0.1.13

Choose a tag to compare

@CaptainCodeUK CaptainCodeUK released this 16 Oct 19:54
· 1 commit to main since this release
ea69469
  • The built-in Storyworks Help story has been rewritten for the current app, with screenshots throughout and new chapters on notes and AI summaries, snippets, search and the trash, exporting, themes and settings, accounts, cloud sync, the Idea Inbox, and getting help.
  • Strikethrough is now Ctrl+Shift+X (Cmd+Shift+X on macOS). Ctrl+Shift+S previously clashed with View → Snippets.
  • Images in the editor now shrink to fit the page instead of overflowing it.
  • Drafts can now be renamed from their tab context menu, with names preserved after reopening the project.
  • Ctrl+Enter submits plot-note forms, Escape cancels, and cancelling a form containing text asks for confirmation before discarding it.
  • Chapter tabs now show drafts, then Notes, then AI Summary when a saved summary exists. Chapter and section Notes glyphs are outlined when empty and filled when they own saved notes; Project Notes opens from the chapter-list header.
  • Plot notes are now editable directly in the Notes tab with Save and Cancel controls. Unsaved text is retained when switching tabs or note scopes.
  • Chapter and section AI summaries now open in read-only AI Summary tabs instead of expandable outline blocks. Section summaries are persisted, and successful generation automatically opens the target's summary tab.
  • AI summaries now take open (not completed) plot notes into account, describing them as planned or upcoming developments rather than events already in the text. Section summaries include the section's notes and those of its chapters.
  • New "Update existing AI summaries" button in the chapter-list header regenerates every chapter and section summary that already exists, without creating new ones.
  • Right-click a chapter or section AI Summary glyph to delete that summary (with confirmation).
  • Fixed: using the bundled local AI model repeatedly (for example, updating many summaries in a row) could eventually crash Workspace. The model now stays loaded between requests, so later requests start much faster, and it is unloaded after a few minutes of inactivity to free memory.
  • Fixed: a longer chat with a character using the bundled local AI model could crash Workspace. The local model now runs in its own background process, so if it does fail, only that request is lost: you see an error and can simply try again. Its working memory is also capped, so it no longer grows to fill your graphics card.
  • When chatting with the bundled local AI model, the message box border now shows whether there's enough graphics memory: green means good to go, amber means it may run out (for example because another program, such as an image generator, keeps grabbing memory), and red means it almost certainly won't fit. Hover over the box for details, including which program is using the memory.
  • If the bundled local AI model crashes because the graphics card ran out of memory, the error now names the program that was using it.
  • Fixed: the bundled local AI model could get stuck repeating itself in a chat until you pressed Stop. Replies are now more varied, avoid repeating the same phrases, and are capped in length as a last resort, with a note if a reply gets cut off.
  • Fixed: after chatting with the bundled local AI model, AI actions such as summaries could fail with a "different Llama instance" error.

Signed builds from Rubber Duck Works. https://rubberduck.works