# Side Chat (Workspace)

**Side Chat** is a powerful LLM-driven assistant integrated directly into your Workspace. It moves beyond simple Q\&A by understanding your entire project context, suggesting code, and even executing tasks autonomously.

<figure><img src="/files/XnsoLW6biVDzHEplbzrZ" alt=""><figcaption></figcaption></figure>

***

## Getting Started

* **Access:** Open Side Chat(<img src="/files/0TiJrt6TE8ArkqBY0xdL" alt="" data-size="line">) from the Sidebar in your Workspace.
* **Model Selection:** Select your preferred LLM from the dropdown menu before starting a conversation. You can check the credit consumption for each model in the [**\[Supported Models\]**](/user-guide/arkain-ai/supported-models.md) guide.
* **Session Context:** Side Chat maintains context within a single session, ensuring a seamless flow of interaction as you build your project.

## The Three Modes of Side Chat

Choose the mode that best fits your current objective. You can switch between them using the dropdown menu in the input box.

#### 1. [Agent Mode](/user-guide/workspace/side-chat-workspace/agent-mode.md) (Autonomous Execution)

* **Best for:** Automating repetitive tasks, creating new files, and running terminal commands.
* **How it works:** An autonomous agent that can independently explore your codebase, modify files, and use terminal tools to complete tasks on your behalf.

#### 2. [Plan Mode](/user-guide/workspace/side-chat-workspace/plan-mode.md) (Collaborative Strategy)

* **Best for:** Designing new features or complex refactorings that require a roadmap.
* **How it works:** The AI interacts with you through interactive questions to clarify your intent. It then generates a structured Action Plan (.md) in your editor. You can view and edit this plan manually.
* **Execution:** Once you are satisfied, click the **\[Approve & Run]** button to transition to Agent Mode and implement the plan.

#### 3. Ask Mode (Read-Only)

* **Best for:** Explaining code, debugging advice, and general questions.
* **How it works:** A conversational assistant that can read your files but cannot modify them. It provides code snippets you can manually copy or apply.

## [Skills](/user-guide/workspace/side-chat-workspace/skills.md) & Capabilities

**Skills** are specialized tools the AI uses to interact with your environment. Powered by the MCP (Model Context Protocol) framework, they allow the AI to perform actions beyond just generating text.

* **Code Search:** Finding specific logic across your entire project.
* **Terminal Execution:** Running builds, tests, or installations.
* **Web Fetch:** Retrieves and references live web content, such as documentation or API specifications, during task execution. Domain permissions and restrictions are managed through the **Preferences** panel.

## Key Features for Precision

#### Adding Context

To get the most accurate results, give the AI visibility into your work:

* **Files:** Click the **\[Add context]** button to reference specific files. You can quickly pick from your 5 most recently opened files or find any file by name or path within the project.
* **Images:** Click the **\[Upload Images]** button or drag and drop images directly into Side Chat. The LLM analyzes the content to enhance its response—perfect for UI debugging or mockup analysis.
  * **Note:** Uploaded images are saved within your project folder. You can freely upload images within your container's storage capacity.

| Feature    | Details                              |
| ---------- | ------------------------------------ |
| Capacity   | Up to 5 images per message           |
| Size Limit | Max 20MB per image (Auto-compressed) |
| Formats    | JPG, PNG, SVG, and more              |

#### Applying Code Suggestions (Ask & Plan Modes)

In **Ask** and **Plan** modes, you can manually integrate suggested code into your editor:

* **Apply/Insert:** Integrate code directly into your editor or at your cursor position with one click.
* **Diff View:** Review changes in a "Before/After" view. Use **\[Accept]** or **\[Discard]** to finalize the edits.

#### Managing Chat History

Never lose a valuable solution by utilizing the built-in history features:

* **Recently Used List:** Quickly load recent conversations from the "Recently used chat" list.
* **Persistent History:** Click the **History icon** (top left) to view all past sessions across your Workspace.
* **Deleting History:** Open the History list, hover over a session, and click the **Trash can icon** to remove it.

#### Message Queue

Side Chat supports a message queue, allowing you to type and send follow-up messages while the AI is still generating a response.

* **Queue messages**: Send up to 5 messages while the AI is responding. Messages appear in order and are sent automatically once the current response completes.
* **Model Switching:** If you change the selected model while messages are queued, the remaining messages will be sent using the newly selected model.
* **Queue Persistence:** Queued messages are saved when you close or refresh the browser tab. When you return to the same conversation, click \[Resume Sending] to continue where you left off, or \[Discard All] to clear the queue.
* **Interruptions:** If you stop a response or an error occurs, the queue pauses. You can click \[Continue Sending] to resume, \[Skip & Continue] to skip the failed message and proceed with the rest, or \[Discard All] to clear the queue.

## Tips for Efficient Use

Maximize your productivity by using the right prompts and context.

* **Be Specific:** Clear instructions yield higher-quality results.
  * *"What’s wrong with this function? \[Code snippet]"*
  * *"Can you optimize this SQL query?"*
  * *"Explain this Python error: \[Error message]"*
* **Vibe Coding:** Describe high-level requirements (e.g., *"Build a simple auth middleware"*), and let Side Chat generate the necessary logic or file structures for you.
* **Keep Context Relevant:** Use \[Add Context] to focus the AI's "attention" on specific files to avoid hallucinations.
* **Review First:** Always use the Diff View to ensure AI suggestions align with your project logic before accepting.
* **Master the Workflow:** For large tasks, start in **Plan Mode** to align on the strategy. Once the plan is solid, use Approve & Run to let the Agent handle the implementation.

***

With Side Chat in Arkain, you have the power of LLMs right at your fingertips to solve problems faster, learn better, and keep track of all your conversations with ease!


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.arkain.io/user-guide/workspace/side-chat-workspace.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.