Give the agent a task in the composer. Add files, quote a previous response, or include a skill when the task needs more context.
Messages can contain up to 120,000 characters. Longer drafts stay in the composer so you can shorten them or split them into several messages.
Attach files
Attach up to eight files per message. Images can be up to 10 MB; other files can be up to 50 MB, subject to the environment’s upload support and limit. The agent receives them on the environment’s machine.
Uploads begin when you add an attachment. All uploads must finish before the message can send. Retry or remove a failed upload. On web and desktop, reloading before an upload finishes requires you to attach that file again.
You can drag or paste images into the web or desktop composer. HEIC and HEIF photos are converted to JPEG there and when selected from the iOS photo library; the image limit applies after conversion. On mobile, you can also send files to VC through another app’s system share sheet.
See images and videos for previewing and saving media.
Queue messages offline on mobile
Mobile keeps local copies of draft attachments, so you can preview them and queue messages while disconnected. Uploads resume when you reconnect. Drafts and queued messages survive app restarts. Signing out of VC Connect keeps that work on your device until you sign back into the same account.
Schedule a message
In a chat, choose Schedule message from the send controls. Send at a date and time you choose, at the provider’s reported quota reset time, or after an agent error. For errors, choose how long to wait and whether to send your draft or retry the previous message with its original attachments. Each schedule runs at most once. An error schedule applies to the current turn and is removed if that turn succeeds or is stopped. Stopping a turn yourself does not count as an error.
On mobile, Schedule task also lets you schedule the first message when creating a task. Choose a date and time or an available quota reset; scheduling after an error requires an existing turn.
Quota scheduling requires an exact reset time for the selected provider account. If the provider does not report one, choose a time yourself. The schedule uses the reported time shown when you save it. Scheduled messages stay visible in the chat, where you can edit, cancel, or send them now. Editing removes the old schedule and returns the message to the composer; save it again when ready.
The environment must be running and connected to the provider when the message is due; your client can be closed. Saved schedules survive environment restarts. If the environment was stopped at the scheduled time, it sends after restarting and becoming available. A scheduled message waits for any running turn to finish.
Switch accounts and recover from limits
Choose another account or model from the thread’s model picker. Confirm the switch; VC stops any running agent first and keeps the thread, history, and workspace. Compatible accounts can resume the provider’s conversation. Other switches start a fresh agent with saved task context, so review the handoff before relying on details outside the saved history. Resolve pending approvals or questions first.
In the thread’s Account and quota recovery dialog, choose your account and policy changes, then select Confirm to apply them. Closing or canceling the dialog discards unconfirmed changes.
When a provider confirms a quota limit, choose Retry now, a reported reset time, a time of your own, or another account. Messages wait while recovery is in progress. If delivery is uncertain after an interruption, review the last messages before retrying. Cancel recovery stops the recovery operation.
Set the environment default under Settings → Providers → Account and quota recovery on web and desktop, or Account and quota recovery in mobile Settings. Choose to ask each time, retry once at a known reset, or try an ordered list of fallback account and model pairs. Project and thread settings can override that default; restoring inheritance uses the containing policy again. Automatic recovery rotates through the list: it starts with the entry after the account that ran out, wraps around to the top, and tries each pair once. It skips an account already known to be limited, including one whose usage meter for the selected model is still full until its reset, unless the provider reports that usage is allowed (for example, with extra usage enabled and budget remaining). List every account, including the one you usually start with, to rotate through all of them. Two entries signed into the same provider login share its quota.
Summarize context when switching providers uses the context compaction model configured in Settings. If summarization is unavailable, VC transfers saved task context directly. Older environments may offer only compatible account switches.
Tasks that may be stalled
VC warns when a running provider or child agent has reported no activity for 10 minutes by default, or 30 minutes while a tool is known to be running. In Settings → General → Task stall detection (or Task stall detection in mobile Settings), you can set each timeout to 1–1440 minutes for each environment or restore the defaults. Changes apply to running tasks within 30 seconds, without restarting the server. Waiting for your approval or answer does not count as a stall. Detection continues while your client is closed; the warning clears when activity resumes.
Silence can also mean slow work. You can keep waiting or choose Stop thread to stop its running work, then send a follow-up to continue the conversation. VC does not automatically retry work in response to this warning.
Custom models
On web and desktop, use Settings → Providers → Models to add an unlisted model with a custom name and options. Only options supported by the provider integration affect turns. Antigravity uses its account catalog and does not support custom models.
Model defaults
VC remembers your provider, model, and model options for new threads. A project’s configured model takes precedence; resetting that project setting returns to the remembered selection.
Leaving reasoning level or service tier unset uses the provider’s own configuration.
Quote an assistant response
On web and desktop, select text within one assistant response and choose Cite in composer. You can add a comment about the quote and write instructions around it.
Select the quote in a draft or sent message to return to its source. If the source is unavailable or has changed, the saved quote remains readable.
Mobile displays saved quotes and comments, but does not create citations or navigate to their sources.
Recall a sent prompt
Press ArrowUp in an empty composer to bring back the last prompt you sent in this thread. Press
ArrowUp again to go further back, and ArrowDown to come forward. Moving forward past the newest
prompt clears the composer. Recall restores the saved message text exactly, including whitespace
and any context text included in that message. Files and images are not reattached. A composer
that holds an attachment or a picked element does not count as empty.
When the composer has text, the arrow keys move the caret as usual. Recall takes over only while
the text is an unedited recalled prompt, with the caret on the first visual line for ArrowUp or
the last visual line for ArrowDown, counting wrapped lines. Editing a recalled prompt turns it
into a normal draft.
Prompt stash
On web and desktop, press Cmd+S on macOS or Ctrl+S on Windows and Linux to save
the current prompt and its attachments for later. Wait for uploads to finish first.
With an empty composer, the same shortcut restores a single stash or opens the
stash menu when there are several.
Stashes containing uploaded files must be restored in their original environment. Those files are retained for 24 hours. After an upload expires, restore the prompt and use Attach again or remove the missing file before sending.
Voice input on iPhone
On supported iPhones with iOS 26 or later, use the composer’s microphone to record, then confirm to transcribe. Text is inserted where your selection was when recording started, ready for you to review and edit before sending.
The first use may download Apple’s speech model and needs a network connection. Later transcription works offline for that language. Recordings can be up to five minutes long. Canceling, leaving the screen, or an audio interruption discards the recording and preserves your existing draft.
Transcription runs on your device. VC deletes the temporary audio after transcription or cancellation; only the message text is sent when you submit.
Commands and skills
Type / for commands or $ to add a skill from the selected environment and
provider. On mobile, both are also available before starting a thread on
New task.
The slash menu also includes skills unless you turn off Settings → General → Show skills in slash menu. Only skills enabled for the provider are listed; turn them on or off from Settings → Skills (Skills).
Provider commands must start the message to run. VC commands such as
/model and /plan, and skill mentions, work on any line.
On web and desktop, Implement in a new thread opens a draft with the proposed plan in a new pane beside the plan, or in place of it when the workspace has no room for another pane, and finishes the original planning task, which no longer waits for your input. Review the context, choose the provider and model, then send when ready to start implementation.
On web and desktop, use Add context in the composer to browse files, skills, MCP tools, or commands, including in a pane’s expanded chat dialog. Use the arrow keys and Enter or Tab to choose a suggestion; Escape closes the list.
Type @project: to reference a project, or @group: to reference a project group
from your sidebar. These suggestions are also included when searching with @.
Selecting one inserts editable names and workspace paths. Projects on another
environment include that environment’s identity; mentioning them does not copy
files or grant the current agent access to that machine.
Type @tool: to find configured MCP servers and recently used tools for the
selected provider. Providers that do not expose their server configuration show
only tools reported in the current conversation, as do connections without access
to server settings. Selecting a server or tool adds an editable request to your
draft; send the message to ask the agent to use it.
Choose Add context → Open VC tool to open Browser, Terminal, or a project tool such as Database or HTTP Client. Project tools use the pane’s workspace and environment; an already open project tool is focused. Browser and Terminal are available after starting the conversation. Opening a tool keeps your message draft.
Send /compact in an existing conversation to reduce context usage when the
provider supports it. Web and desktop also offer compaction from the context meter.
Images and videos in messages
Select an image or video attachment or link to preview it. Playback support depends on your browser or device; save an unsupported video to open it in another app.
On web and desktop, right-click media to save it or copy its path or URL. On mobile, touch and hold an image or video thumbnail and choose Save or share. On iOS, return to the thumbnail to open this menu after watching a full-screen video.
File links refer to the environment’s machine, including when you connect remotely. Previews use the original file, even outside the workspace. Moving or deleting it can break the preview, so save a copy if you need to keep it.
Edit workspace files
On web and desktop, open a text file in Files and choose Edit to focus its
source. Changes save automatically to the connected environment. Use Save or
Cmd+S / Ctrl+S in the source editor to save immediately. If saving fails, the
draft stays visible; choose Retry save after restoring the connection.
Markdown and HTML can switch between source and preview. Files larger than the 1 MB preview limit, media, attachments, and files outside the workspace are read-only.
Files outside the workspace
Follow an agent’s file link to read a report or other file outside the workspace. These files open read-only. An HTML file outside the workspace cannot load scripts, styles, or images from neighboring files.
HTML and PDF files in the file viewer
On web and desktop, HTML and PDF files open as rendered pages. Switch an HTML file to source view to read its markup; a link to a specific line opens source automatically. HTML previews cannot access your VC session.
On mobile, select a PDF attachment or link to open it. iOS uses the native viewer; Android opens the system chooser.
Work with multiple threads
On desktop and web, the plus in the workspace toolbar adds a chat pane beside the current one. The new pane opens a draft with the source pane’s project, checkout, model, and working modes. Change the provider or model from the composer whenever you want. Each pane has its own thread and composer draft. Selecting a thread that is already open focuses its existing pane.
Messages you send from a pane while its agent is working wait in that thread’s queue instead of steering the running turn. They start in order as each turn finishes; the pane header shows how many are waiting, and you can remove a queued message from the timeline before it starts. Once a thread has queued messages, ordinary sends from any device join the end of the queue. Press Ctrl+Enter on macOS, Windows, or Linux to send directly to the current chat and bypass the queue. Cmd+Enter also works on macOS. Shift+Enter adds a new line.
Adding or closing a pane divides the width equally among the panes in its row. If a split makes panes too narrow or too short, the window arranges them into a balanced grid. Use Arrange panes in the workspace toolbar to arrange them yourself. Drag a divider to resize panes, or focus it and use the arrow keys. You can maximize a pane and restore the layout. Drag the title or empty area of a pane header onto another pane to swap their positions, or onto that pane’s left or right edge to split into columns (top or bottom edge for rows); press Escape to cancel. With the pane title focused, Alt+Left/Right swaps with the previous or next pane. Closing a pane does not delete its thread or stop a running agent. Up to 32 panes can be open in one window. The pane arrangement and the thread in each pane are saved with that workspace in this browser. Creating a workspace with the project plus button or New workspace leaves the current panes where they are and opens one new pane in the new workspace. Selecting a workspace restores only its panes. Use Reset workspace panes in the command palette to collapse the current workspace back to a single pane.
When a pane is too small for the composer, or three or more panes are stacked in its column, use its chat icon to open a larger message dialog with the model picker and prompt editor. Closing the dialog keeps your draft. Enlarging or unstacking the pane restores the inline composer.