Secondbrain

<p align="center"> <img src="https://raw.githubusercontent.com/lucide-icons/lucide/main/icons/brain.svg" width="80" height="80" alt="NeuroCanvas Logo" /> </p> <h1 align="center">🧠 NeuroCanvas</h1> <p align="center"> <strong>Visual AI canvas for prompt engineering and knowledge graphs</strong> </p> <p align="center"> <a href="#-download">Download</a> • <a href="#-features">Features</a> • <a href="#-file-manager">File Manager</a> • <a href="#-attachments-files-in-cards">Attachments</a> • <a href="#-quick-start">Quick Start</a> • <a href="#-hotkeys">Hotkeys</a> • <a href="#-api-setup">API Setup</a> </p> <p align="center"> <img alt="Next.js" src="https://img.shields.io/badge/Next.js-14-black?style=for-the-badge&logo=next.js" /> <img alt="React" src="https://img.shields.io/badge/React-18-61DAFB?style=for-the-badge&logo=react&logoColor=white" /> <img alt="Electron" src="https://img.shields.io/badge/Electron-39-47848F?style=for-the-badge&logo=electron&logoColor=white" /> <img alt="TypeScript" src="https://img.shields.io/badge/TypeScript-5-3178C6?style=for-the-badge&logo=typescript&logoColor=white" /> </p>

---

📥 Download

Download the latest version for Windows:

• Installer (.exe)
• Portable Version

---

🎯 What is it?

NeuroCanvas is an innovative tool for visually building AI prompts on an infinite canvas. Instead of a linear chat, you create a knowledge graph where each card connects to others, and context is passed automatically.

💡 The Problem

Traditional AI chats suffer from a critical issue: linear history. When exploring a complex topic, you inevitably drift into side branches, making it painful to return to the main conversation line.

✨ The Solution

NeuroCanvas allows you to:

• Branch dialogues like Git branches
• Quote specific parts of answers to continue
• Visualize connections between ideas
• Search across all canvases using hybrid AI search

---

🌟 Features

🤖 Multi-AI Support

• Curated Catalog: Built-in support for top-tier models (GPT-5.2, Claude Opus 4.5, Gemini 2.5 Flash, DeepSeek V3.2, etc.).
• OpenRouter & Custom: Full compatibility with OpenRouter and any OpenAI-compatible API.
• Smart Configuration: Automatic handling of context window limits and model capabilities.

🧠 System Instructions

Define global system instructions and behavioral rules for the LLM that apply to the entire canvas context. Control the persona, tone, and constraints of your AI assistant.

🎨 Visual Canvas

• Infinite canvas with navigation and zoom
• Drag-to-Create: drag a connection to create a card
• Smart Selection: intelligent selection and bulk operations
• Dynamic Positioning: new child cards automatically position to the right of parent with proper spacing
• Resizable Cards: drag the right edge to adjust card width
• Automatic Context: parent cards form the context for children
• Colored Connections: Visual distinction between standard parent-child links and quote-based connections.

📖 Reading Mode

Focus on the content with a distraction-free reading mode for any card. Ideal for studying long responses or notes without UI clutter.

📝 Card Types

• AI Cards: interact with LLM, generate responses, support context inheritance
• Note Cards: personal notes with double-click to edit, formatted as context for AI
  • Title and content stored separately
  • Auto-summarization for context compression
  • Can be quoted and referenced like AI cards

💬 Quote System

• Quote from Response: select text from any card's response to create a linked card
• Context Preservation: quoted text is included in context along with full response/summary
• Quote Invalidation: automatic detection when source response changes
• Quote Update: re-select quote when source is modified
• Visual Highlighting: quoted sections highlighted in context viewer
• Colored Links: Connections created from quotes have distinct colors on the canvas.

🧠 NeuroSearch (RAG)

NeuroSearch is a powerful feature that transforms standard semantic search into an active part of your thinking process. It allows you to connect any card to contextually similar information from your entire knowledge base without creating direct links.

• Semantic Connection: Click the Brain icon 🧠 to instantly find and connect relevant cards based on meaning, not just keywords.
• Virtual Relationships: Connected cards act as "virtual parents," providing their full context to the current card.
• Context Inheritance: Descendant cards inherit NeuroSearch results as "virtual grandparents" (summarized), understanding the broader context of their ancestors.
• Stale Detection: The brain icon turns orange 🟠 if any connected card is updated, signaling that your context might be outdated. Click it again to refresh the search.
• Granular Control: Manually exclude specific search results from the context if they aren't relevant.

🔍 Smart Search (RRF)

Combines 4 search methods for perfect results:

1. BM25 (Keyword matching)
2. Semantic Search (Vector search by meaning)
3. Fuzzy Search (Typo-tolerant search)
4. Exact Match (Precise phrase matching)

Enhanced Navigation:

• Keyboard Navigation: arrow keys to navigate results
• Auto-scroll: selected result stays in view
• Visual Highlighting: clear indication of selected item
• Quick Access: Ctrl+P or Ctrl+Л (Russian layout)

🎯 Context Management

• Context Viewer: see full context hierarchy sent to LLM
• Exclude from Context: manually exclude specific ancestor cards
• Expand/Collapse All: quickly manage context visibility
• Smart Context Types:
  • Full response for direct parents
  • Quotes for referenced sections
  • Summaries for distant ancestors

📂 File Manager

Built-in global library for your knowledge files:

• Upload & Organize: folders, move/rename, replace file contents.
• Fast Preview: open text and images instantly (no external apps required).
• Attach to Cards: use files as context for AI generation.
• Safe Deletion: Trash, Restore, Empty Trash, and Garbage Collection (GC).
• Usage Awareness: see which canvases/cards use a file and jump there.
• Background AI Analysis: generate summaries (text) and descriptions (images) for better context.

⚙️ Advanced Capabilities

• Embeddings Model Selection: tune search quality (text-embedding-3, multilingual-e5, etc.)
• Corporate Mode: work in networks with SSL inspection
• Summarization: automatic compression of long contexts
• Local Data: all canvases/search index are stored only on your device (API keys are not stored in localStorage)
• Undo/Redo: full history with Ctrl+Z / Ctrl+Y
• Batch Operations: mass collapse/expand for selected cards
• Stale Detection: automatic tracking of outdated cards when context changes

---

📂 File Manager

NeuroCanvas includes a File Manager (right sidebar) — a global library where you keep documents and assets that can be reused across your canvases.

What you can do

• Upload files

  • Drag & drop files into the File Manager.
  • Or click to select files from your system.
  • Uploading supports name conflicts (same name in the same folder) and lets you choose:
    • Replace the existing file (keeps the same file entry),
    • Upload as a new file (with a new name),
    • Skip.

• Browse & organize

  • Folders with nested structure.
  • Move files between folders.
  • Rename files and folders.
  • Replace a file’s contents while keeping it as the same library item (useful when you update a document).

• Preview & download

  • Preview text documents directly in the app.
  • Preview images instantly.
  • Download any file from the library.

• Find files quickly

  • Search by file name.
  • Optional filters:
    • filter by canvas (show files used on a specific canvas),
    • filter by file extension (e.g. md, pdf, png).

• Trash & cleanup

  • Trash moves a file into the library Trash.
  • Restore brings it back.
  • Empty Trash permanently removes files in Trash.
  • GC (Garbage Collection) permanently removes files that are safe to delete (advanced cleanup).

• See where a file is used

  • The File Manager can show where a file is referenced (by canvas/card).
  • You can jump directly to the card that uses a file.

Tabs: Files vs Trash

• Files tab: your normal library view (folders, filters, upload).
• Trash tab: a global list of deleted files (no filters, to keep restore/cleanup simple and predictable).

---

📎 Attachments (Files in Cards)

Attachments are how you bring files into your AI prompts. A card stores a reference to a library file, and uses that file as context during generation.

Attach a file to a card

You have two convenient ways:

• Attach from your computer (upload + attach)

  • Click the paperclip on a card and select files.
  • Files are uploaded to the library and immediately attached to the card.

• Attach from the File Manager (drag & drop link)

  • Drag a file from File Manager onto a card.
  • This attaches the existing library file without re-uploading.

Preview attachments

• Click an attachment thumbnail in the card to preview it:
  • images open instantly,
  • text files are loaded on demand and shown as plain text.

Include / exclude attachments from context (per card)

Sometimes a file is attached but you don’t want it to influence a sp