> ## Documentation Index
> Fetch the complete documentation index at: https://documentation.orbitdev.org/llms.txt
> Use this file to discover all available pages before exploring further.

# Chat and manage conversation history in Orbit AI

> Learn how to start or continue a chat, use attachments, voice, image and artifact generation, stop generation, and find or manage conversations from your history.

Orbit AI lets you start new conversations and pick up existing threads from the main app. You can attach files, use voice input, generate images and documents, and manage your past chats from the history page. Conversations are addressable by URL, so you can open a specific chat directly or share its link.

## Chat screen controls

The main chat screen includes several controls in the header and toolbar. Some may be hidden on narrow screens or when a feature is unavailable.

| Control | What it does |
| - | - |
| **Mode selector** | Choose between Chat, Fast, Build, and ProPulsar. A lock indicator appears when your account cannot use a mode. |
| **Model picker** | Choose the model for the current work. Options depend on your workspace, plan, and entitlements. |
| **Extra chat** | Open an additional chat pane where supported. |
| **Widgets** | Display configurable chat widgets when your account has widget access. |
| **New conversation** | Start a fresh conversation. |
| **Rename conversation** | Change the title of the current saved conversation. |
| **Delete conversation** | Start the app's conversation removal action. Check the confirmation if shown. |
| **Composer** | Enter a prompt, attach supported input, use voice features, and submit. |

<Note>
  On mobile, some actions appear in a different layout or are omitted.
</Note>

## Start or continue a chat

<Steps>
  <Step title="Open the app">
    Go to [https://www.app.orbitdev.org](https://www.app.orbitdev.org) or [https://www.app.orbitdev.org/app](https://www.app.orbitdev.org/app) while signed in. The main chat view is the default landing page.
  </Step>

  <Step title="Send a message">
    Type your prompt in the composer and press **Send**. The app creates a new thread automatically if you are not continuing an existing one.
  </Step>

  <Step title="Continue an existing thread">
    If you are already inside a conversation, new messages are appended to that thread. To start a fresh chat, use **New conversation**.
  </Step>
</Steps>

## Conversation routes

The app supports direct conversation URLs. You can navigate to a specific chat using these routes:

| Route | Purpose |
| - | - |
| `/chat` | Chat entry and chat sessions |
| `/chat/new` | Start a new conversation |
| `/chat/:conversationId` | Open an existing conversation by its ID |

You can open a conversation directly from History, or navigate to `/chat/:conversationId` if you know the conversation ID.

## Model selection

The chat home includes a model picker that loads available options from the server. Your choices are filtered by your workspace, plan, and any access restrictions configured for your account. Some models or capabilities may require an eligible plan or access code.

## Generation status and partial output

While the assistant is generating a response, the UI shows live status and partial output from persisted generation progress. The display is updated from real generation state, not simulated token streaming.

## Stop generation

You can stop an in-progress generation at any time. The partial output produced up to that point is saved to the conversation.

## Writing useful prompts

Give Orbit enough information to understand the job. A useful prompt usually includes:

1. **Goal**: what you want made, explained, planned, or changed.
2. **Context**: relevant audience, platform, source material, or background.
3. **Constraints**: length, tone, tools, deadline, or things to avoid.
4. **Output shape**: checklist, table, step-by-step guide, code, or another format.
5. **Success criteria**: how you will tell whether the result is complete.

For example, instead of "help with my project," try: "Create a launch checklist for a small portfolio website. Group it by design, implementation, testing, and publishing. Keep each item actionable and flag anything that requires an account or paid service."

<Warning>
  Do not include confidential information you do not want processed by an AI provider.
</Warning>

## Continue, branch, or start fresh

Continue in the same conversation when the next request depends on earlier answers. Start a new conversation when you want a clean context, are changing topics, or want a separate history item. Use **New conversation** rather than deleting the old one if you simply want a fresh workspace.

If the first result is not right, explain what to change and what should remain unchanged. Ask for one correction at a time when it is important to preserve parts of a prior response.

## Attachments and media

The composer supports several attachment and media types:

* **File attachments**: upload files to include as context with your message. Supported formats and size limits depend on your workspace and model configuration.
* **Image generation**: generate images directly from the chat interface when your plan and model support it.
* **Voice input**: record and send voice messages. This uses your browser or device media capability, so microphone permission may be required.
* **Artifacts and documents**: generate structured documents and artifacts from within the chat.

Before uploading, check the following:

* Confirm you have the right to share the file.
* Remove unrelated personal data where practical.
* Do not upload passwords, recovery phrases, private keys, or sensitive payment details.
* Keep a local original if the file is important.
* Check the final prompt so it is clear what Orbit should do with the file.

<Note>
  Exact file size limits, supported MIME types, voice transcription behavior, and generation availability vary by platform and configuration.
</Note>

## Voice session controls

Voice controls may start a voice session or accept speech, depending on the UI and platform. The browser or operating system may ask for microphone permission. Grant permission only when you intend to use voice, and revoke it in browser or device settings if you no longer want microphone access.

In a voice session, use the visible mute, retry, stop or end, or close controls as appropriate. A transcript may be shown.

<Warning>
  Check what was transcribed before relying on it. Names, numbers, accents, and background speech can be misheard. Voice features may not behave identically in web, desktop, and mobile builds.
</Warning>

## Find and manage conversations

Your chat threads are stored remotely and accessible from the history page.

<Steps>
  <Step title="Open history">
    Navigate to [https://www.app.orbitdev.org/history](https://www.app.orbitdev.org/history) while signed in.
  </Step>

  <Step title="Search">
    Use the search field to filter your conversation list by content or title. History and Library searches cover chat threads and library items or assets. They are not a full-text search over every individual message.
  </Step>

  <Step title="Open a thread">
    Select any conversation to load its messages in the main chat view.
  </Step>

  <Step title="Rename">
    Choose **Rename** from the conversation row to update its title.
  </Step>

  <Step title="Remove">
    Choose **Remove** from the conversation row to delete it from your active history. Removal may be unrecoverable, so treat this action as permanent.
  </Step>
</Steps>

## Library details

Open **Search & Library** or `/history` to see a combined view of conversation history and saved library items or assets, when available.

* Use the **Search library** field to search available item titles and related indexed text.
* Choose a filter: **Everything**, **Chats**, **Assets**, **Images**, **Files**, **Projects**, **Agents**, or **Automations**.
* Sort by **Last updated**, **Date created**, or **Name**.
* Open a chat card to resume that conversation. Open an asset or other item card to inspect its details.

Library cards can display favorite or pinned markers. Other row actions can depend on item type and current build. Archived or trashed records may be excluded from the default list.

<Warning>
  Treat deletion as irreversible unless the UI explicitly offers a restore action. Do not use deletion as a way to hide an item if you may need it later. Preserve important work elsewhere.
</Warning>

## Prompt safety

<Warning>
  Do not include passwords, payment card details, private access codes, or confidential information in prompts. AI responses can be inaccurate, so verify important facts and review generated code or documents before using them.
</Warning>

## Signed-out preview

When you are not signed in, the chat home shows a local preview with canned responses. These are not real AI outputs. Sign in to get live model responses and cloud persistence.

## Troubleshooting

If a message fails to send, check your network connection and retry once. Persistent errors may indicate a provider or workspace configuration issue. If history does not load, confirm you are signed in and refresh the page.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.