Brain Manager — local AI context studio

Library

Brains live in this browser's local storage. Export a file to move one elsewhere.

Revisions

Every save snapshots the brain with a timestamp — stored only in this browser. One click loads a snapshot back into the editor.

Editor

Brain name Persona — who this brain is (system prompt) Memory anchors — one per line (facts this brain always carries) Formatting directives — one per line Variables — KEY=value, one per line. Use {{KEY}} anywhere above to substitute.

Google Gemini

OpenAI Chat

Vertex AI REST

Estimated tokens: —

Gemini context window: 1,000,000 tokens

GPT-class context window: 131,072 tokens

Payload preview only — it is not sent anywhere. Token counts are a heuristic estimate (≈ characters ÷ 4), not exact tokenizer output.

Live test — real provider call

Sends the current editor brain (persona + memory anchors + directives, with {{variables}} substituted) as the system prompt and your message as the user turn — as a real HTTPS call from this browser straight to the provider's API. This page is a static file with no backend; nothing of yours touches any server of ours.

Provider API key — typed into your browser, never leaves it except to the provider Remember this key in this browser (localStorage only)

Test message

Key handling, stated plainly: the key you type is held only by this page. If you tick "remember", it is saved in this browser's localStorage and can be forgotten with one click. On send it goes only to the provider you chose — generativelanguage.googleapis.com for Gemini, api.openai.com for OpenAI. Provider errors (bad key, quota, billing) are shown back to you verbatim — never dressed up as success.

Raw provider JSON (click to expand)