> ## Documentation Index
> Fetch the complete documentation index at: https://012e27a-docs-2.manicule.dev/llms.txt
> Use this file to discover all available pages before exploring further.
> For every page in one file, fetch https://012e27a-docs-2.manicule.dev/llms-full.txt (large).
> To search and read these docs as tools, connect the MCP server at https://012e27a-docs-2.manicule.dev/api/mcp.
> Every agent surface: https://012e27a-docs-2.manicule.dev/.well-known/llms-manifest.json

# Desktop Workspace Overview (/desktop/overview)

<!-- agent-signals: reading_time_min: 2 · est_tokens: 1002 · updated: 2026-11-12 -->
Related: [Agentic Canvas Mode](/desktop/agentic-mode.md), [Code Editor & Highlight-to-Ask AI](/desktop/code-editor.md), [Multi-File Visual Diff Viewer](/desktop/diff-viewer.md), [Transparency Dock & Approval Gates](/desktop/transparency.md), [Integrated PTY Terminal](/desktop/terminal.md), [Cloud Studio Quickstart](/getting-started/quickstart.md)

> Architecture, performance benchmarks, and capabilities of the native BhaiCode desktop app.

The **BhaiCode Desktop App** is a native development workspace built with **Tauri 2, Rust,
and React 19**. It gives you high-velocity AI assistance with local-first privacy, native
terminal integration, and a small memory footprint.

Rather than bundling Chromium and Node.js runtimes inside Electron, BhaiCode Desktop renders
with your operating system's native webview and runs filesystem, process orchestration, and
PTY terminal work through a compiled Rust core.

<CardGroup cols="3">
<Card title="~7–8 MB" icon="package">Installer size, versus ~120–150 MB for a typical Electron editor.</Card>

<Card title="~45 MB RAM" icon="memory-stick">Idle memory footprint, versus ~500–800 MB.</Card>

<Card title="< 350 ms" icon="timer">Cold startup time, versus 2.5–4.5 s for a typical Electron editor.</Card>
</CardGroup>

## Architecture

```mermaid
flowchart TD
    subgraph F["React 19 frontend"]
        CANVAS["Agentic Canvas<br/>plan timeline"]
        EDITOR["Editor mode<br/>CodeMirror 6"]
        DIFF["Diff viewer<br/>hunk review"]
        TERM["PTY terminal<br/>xterm.js"]
    end
    subgraph R["Tauri 2 Rust core"]
        IPC["Typed IPC bridge"]
        FS["Filesystem<br/>denylist guards"]
        PTY["portable-pty<br/>process manager"]
        KEY["OS keychain"]
    end
    subgraph A["Agent swarm"]
        BRAIN["Bhai · Planner · Executor"]
    end
    CANVAS --> IPC
    EDITOR --> IPC
    DIFF --> IPC
    TERM --> IPC
    IPC --> FS
    IPC --> PTY
    IPC --> KEY
    IPC --> BRAIN
```

### Performance versus a typical Electron editor

| Metric                | BhaiCode Desktop (Tauri 2)         | Typical Electron editor |
| --------------------- | ---------------------------------- | ----------------------- |
| Installer size        | \~7–8 MB                           | \~120–150 MB            |
| Idle memory footprint | \~45 MB RAM                        | \~500–800 MB RAM        |
| Cold startup time     | \< 350 ms                          | 2.5–4.5 s               |
| Execution security    | Native OS keychain & path denylist | Plaintext config files  |

## Two working modes

<Tabs>
  <Tab title="Agentic Canvas">
    A focused interface for task delegation (`Cmd + 1` / `Ctrl + 1`):

    * Step-by-step implementation plan progression.
    * Subagent delegation, e.g. system design to **Planner** or heavy refactors to **Executor**.
    * Real-time tool execution logs.
    * Permission approval gates for terminal commands and sensitive actions.

    [Learn about Agentic Canvas Mode →](/desktop/agentic-mode)
  </Tab>

  <Tab title="Workspace Editor">
    A dual-pane developer environment (`Cmd + 2` / `Ctrl + 2`):

    * File explorer with Git status indicators.
    * **CodeMirror 6** editor with syntax highlighting for 100+ languages.
    * **Ask AI on Selection**: highlight code to explain, refactor, or test it inline.
    * Integrated **PTY terminal** running your system's native shell.

    [Open the editor guide →](/desktop/code-editor)
  </Tab>
</Tabs>

## Local-first security and BYOK

* **Zero remote code storage**: your source stays on your local filesystem.
* **Native OS keychain vault**: keys for OpenAI, Anthropic, Gemini, and Nous Portal are stored
  in **macOS Keychain**, **Windows Credential Manager**, or **Linux Secret Service**, never in
  unencrypted config JSON.
* **Direct provider connection**: requests go from your machine to the provider endpoint
  without a BhaiCode proxy.
* **Air-gapped offline models**: connect local models via Ollama at `http://localhost:11434`.

<CardGroup cols="2">
<Card title="Review every change" icon="git-compare" href="/desktop/diff-viewer">Multi-file diff review with hunk-by-hunk accept/reject and checkpoint rollback.</Card>

<Card title="Stay in control" icon="shield-check" href="/desktop/transparency">Risk tiers decide what runs automatically and what waits for your approval.</Card>
</CardGroup>