Brainstorming Partner

UPDATE v2.0.0

New: Plan Manager integration

I have a plan manager script I am not sure if I ever shared. I did some integrations to the brainstorming script with the plan manager. For now, I will share the plan manager script here but will eventually create a separate chat for it. The plan manager and the brainstorming scripts can work without each other's fine.
If you have the Plan Manager script installed, a Create PM Plan button appears in the Brainstorm tab. The AI writes a full structured plan based on your chat discussion and story context, you review and edit it right there in the sidebar, then submit it straight to Plan Manager. no copy-pasting. If a plan already exists, the button becomes Update PM Plan instead: it pulls the live plan back in so the AI revises it rather than blindly replacing it. Long plans can be extended with a Continue button if they get cut off.
(Don't have Plan Manager? Nothing changes. The button simply won't appear.)

New: Continue button for cut-off replies

When the AI's answer gets cut off by the output limit, a Continue reply button now appears below it and picks up exactly where it stopped. No more losing the end of a long answer.

New: Clear Chat is no longer a one-click disaster

Clear Chat now asks you to confirm first. On top of that, cleared chats are saved to a new chat archive in the History tab (the last 15 are kept) — you can preview any of them and restore them as your live chat. Archived chats cost you nothing: they're never sent to the AI.

Improved: smarter token handling with generation helpers

This is a bit of a technical update you may not notice any difference.
The script now cycles through your generation sources (main pool → Helper 1 → Helper 2 → … → back to main) and stays on whichever one is working, instead of retrying an empty pool every single message. It remembers its position across reloads, and it can now recover on its own if a helper gets stuck mid-generation.

Fixed: the AI writing fake "entry created" messages

When you created or updated a lorebook entry, the confirmation line was being stored as though the AI had said it — so the AI would start imitating them and announce entries it hadn't actually made. These are now proper script notices: visually distinct, clearly labeled, and correctly understood by the AI. Old chats are cleaned up automatically on load.

Fixed: lorebook entries ignoring your formatting requests

If you asked for bullet points, labeled sections, or "convert this to prose," the AI would often ignore it and default to its own format. Your formatting request is now respected — whether you said it in the focus prompt or earlier in the conversation.

Also updated: BP — Generation Helper (v1.2.0)

If you use generation helpers, please update them alongside the main script.

Fixed: helpers getting permanently stuck.

When a helper ran out of tokens, it could hang forever without ever reporting back — and stay unusable until you reloaded. It now times out, marks itself as spent, and hands off to the next source.

Fixed: replies vanishing in transit.

A helper could finish generating and the result would silently never arrive. The response is now trimmed down to just what's needed before being sent back.

Improved: "Reset Helpers"

now properly unsticks a jammed helper instead of only clearing its exhausted flag.

New: Flip chat order

A new "⇅ Oldest first / ⇅ Newest first" button at the top right lets you toggle between viewing the chat oldest-to-newest (default) or newest-to-oldest. When flipped to newest-first, the message input box automatically moves to the top so it stays next to the most recent message instead of being buried at the bottom of a long conversation. Your preference is saved per story.

New: Tabs always visible while scrolling

The Brainstorm / Scenario / History tab row now stays pinned at the top of the panel even when you scroll down through long chat history, so you can switch tabs anytime without scrolling back up. The tab buttons are also smaller and tighter so they don't take up much space.

Fix: Token counter now reflects what's actually sent

The "Story" token counter at the bottom of the Brainstorm tab used to only count the raw story text. It now correctly includes Memory, Chapter Summaries, and pinned Lorebook entries — all of which the chat AI actually sees alongside the story. If you suddenly notice a higher Story count than before, that's not a regression — the counter is finally telling the truth about what's been getting sent.

Improved: Short-term Scenario Guide is much tighter

The short-term Scenario Guide was producing chunky bulleted blocks that felt heavy in the author's note. The prompt has been rewritten to produce a tight 1–3 line nudge instead (max ~60 words total), so it steers the immediate next scene without crowding out anything else you have in your author's note. Existing short-term guides will tighten up automatically on the next refresh.

Initial release