> ## 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

# Agentic Canvas Mode (/desktop/agentic-mode)

<!-- agent-signals: reading_time_min: 2 · est_tokens: 629 · updated: 2026-11-12 -->
Related: [Desktop Workspace Overview](/desktop/overview.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)

> Delegate multi-step tasks to Bhai, follow the plan live, and review every change.

**Agentic Canvas Mode** is the desktop workspace's task delegation mode. You set the goal and
the constraints; the agent swarm plans the architecture, edits across files, and validates the
result before handing it back.

Toggle it with `Cmd + 1` / `Ctrl + 1`; switch to the editor with `Cmd + 2` / `Ctrl + 2`.

## Execution lifecycle

```mermaid
flowchart LR
    A["1. Prompt"] --> B["2. Scope analysis"]
    B --> C["3. Planner<br/>checklist"]
    C --> D["4. Steps<br/>edits and approvals"]
    D --> E["5. Multi-file<br/>diff review"]
```

<Steps>
<Step title="Describe the task">Write the goal in English, Hindi, Hinglish, or another supported language. Bhai inspects
your file tree and dependency graph to identify affected modules.</Step>

<Step title="Plan before code">For complex or multi-file work, Bhai asks the **Planner** for an ordered checklist
before touching a file.</Step>

<Step title="Execute step by step">Bhai reads files and symbol references, applies surgical edits, and runs your compiler,
linter, or test commands in the terminal.</Step>

<Step title="Approve risky steps">Destructive commands — deleting files, installing packages, running external scripts —
pause in the [Transparency Dock](/desktop/transparency) until you decide.</Step>

<Step title="Review the diff">When the plan completes, Bhai presents one consolidated visual diff of every change for
approval in the [diff viewer](/desktop/diff-viewer).</Step>
</Steps>

## What you see while it works

| Panel              | Shows                                                           |
| ------------------ | --------------------------------------------------------------- |
| **Plan checklist** | Each step with `Pending`, `Running`, `Done`, or `Failed` status |
| **Thought stream** | The agent's reasoning, kept separate from the final code        |
| **Tool log**       | File reads, searches, edits, and commands as they run           |
| **Diff cards**     | Per-turn changes, inspectable without leaving the conversation  |

### Living task checkpoints

If a compiler or test error appears, Bhai creates a corrective sub-task and fixes it before
moving on. Nothing is applied silently — every change lands in the diff for review.

<Info title="Reverting a session">Every completed task records a Git checkpoint. Use **Rollback to checkpoint** in the diff
toolbar to restore the working tree to its state before the prompt.</Info>