Skip to main content

Slash Commands

Slash commands are typed into the ByteAsk TUI composer. Press / and an autocomplete menu appears; start typing to filter, then press Enter or Tab to run. Commands marked during task work while the agent is actively running; the rest run between turns.

tip

The commands shown in autocomplete depend on your account. The ByteAsk account commands below (/usage, /referral-code, /voucher, /upgrade) appear when you're signed in to a ByteAsk account with a magic link.


ByteAsk account, usage & billing

These commands are specific to ByteAsk. They manage your plan, your usage limits, and your account credit. All four work during a task.

/usage - see your plan and remaining limits

Shows your current plan and how much of each usage window you've spent. Usage is metered over two rolling-anchored windows: a 5-hour window and a weekly window. The card shows a percentage bar and the exact reset time for each.

/usage

Example output:

ByteAsk usage
Plan: Free
5-hour [███░░░░░░░] 28% resets in 3h 47m (6:12 PM)
Weekly [█████░░░░░] 51% resets in 5d 19h (Sat Jul 11, 9:30 AM)
Extra no credit yet - redeem one with /voucher <code>
Upgrade: type /upgrade (Pro $20 · Max $100)

The plan rows are shown as percentages only. The one line that shows dollars is the Extra row - your account credit (see /voucher and /referral-code). Once you have credit it reads like:

Extra [██░░░░░░░░] $2.10 / $10.00 used ($7.90 left)
note

Credit ("Extra") is spent automatically after your plan's 5-hour and weekly budgets are used up, so it extends your usage past the plan limit rather than replacing it.

/referral-code - earn credit by referring

Prints your personal referral code plus ready-to-share install commands for every install method. When someone installs ByteAsk with your code and signs up, you both get $10 of credit.

/referral-code

Example output:

Your ByteAsk referral code
4EXQTXUK
Share any of these - you and your friend both get $10:
curl curl -fsSL https://code.byteask.ai/install.sh | BYTEASK_REF=4EXQTXUK sh
pip pip install byteask && BYTEASK_REF=4EXQTXUK byteask
npm npm install -g @byteask/cli && BYTEASK_REF=4EXQTXUK byteask

The $10 for each side lands when the person you referred signs in for the first time. It appears as Extra in /usage. There's a lifetime cap on how many referrals earn credit.

/voucher - redeem account credit

Redeem a voucher code for a one-time credit top-up. Credit covers usage past your plan limits.

/voucher <code>

Running /voucher with no code prints the usage hint. On success:

Voucher LAUNCH25 redeemed
Added $25.00 of credit - your balance is now $25.00.
It shows as "Extra" in /usage and covers spend past your plan limits.

Each voucher can be redeemed once per account. A code that's already been used on your account reports "already used" rather than topping up again.

/upgrade - move to Pro or Max

Upgrade your plan for more usage. Running /upgrade with no argument lists the options:

/upgrade
Upgrade your ByteAsk plan
Pro - $20/month → type /upgrade pro
Max - $100/month → type /upgrade max
Your plan activates automatically right after payment.

Pick a plan to get a checkout link:

/upgrade pro
/upgrade max
Upgrade to Pro - $20/month
https://rzp.io/i/xxxxxxx
Open the link to pay; your plan upgrades automatically after payment.

Your plan is upgraded automatically once payment completes - no need to sign in again.

/login and /logout - manage your session

/login signs in (or switches account) and /logout signs out. Because the sign-in token lives in the byteask wrapper (not the engine), these commands write a small marker file and exit the TUI; the wrapper then completes the magic-link sign-in (or clears the token) and relaunches you.

/login
/logout

Model and behavior

CommandDuring taskDescription
/modelyesOpen the model picker. Choose a cloud model (or a self/<alias> self-hosted model) and, for models that support it, a reasoning effort.
/permissionsyesSet what ByteAsk is allowed to do (read files, run shell commands, write files) without restarting.
/approveyesApprove one retry of a recent auto-review denial.
/plan-Switch to Plan mode: the agent drafts a step-by-step plan for you to review before it executes.
/personalityyesChoose the agent's communication style.
/experimental-Toggle experimental features for the current session.

Conversation flow

CommandDuring taskDescription
/new-Start a new conversation without exiting. The current session is saved.
/compact-Summarize the conversation to free up context window space in long sessions.
/clear-Clear the terminal display and start a fresh conversation.
/copyyesCopy the last model response as Markdown to the clipboard.
/rawyesToggle raw scrollback mode (unrendered response text).
/diffyesShow a git diff of all changes made this session, including untracked files.
/mentionyesInsert a reference to a file into the composer.
/goalyesSet or view the goal for a long-running task.
/renameyesRename the current session thread.

Session management

CommandDuring taskDescription
/resume-Open the session picker to resume a saved conversation.
/fork-Fork the current session from its present state.
/archive-Archive this session and exit (hidden from the default list, still resumable).
/delete-Permanently delete this session and exit.

Multi-agent

CommandDuring taskDescription
/agentyesSwitch the active agent thread.
/subagentsyesList all active agent threads.
/sideyesStart an ephemeral side conversation to ask a quick question without interrupting the main task.
/btwyesAlias for /side.

Tooling

CommandDuring taskDescription
/mcpyesList connected MCP servers and their tools. Use /mcp verbose for full tool schemas.
/appsyesManage installed ByteAsk apps.
/pluginsyesBrowse and manage plugins.
/skillsyesUse skills (reusable task templates) to improve performance on specific task types.
/hooksyesView and manage lifecycle hooks.
/memories-View, add, edit, or delete memories carried across sessions.
/import-Import setup, project instructions, or chat history from Claude Code.
/init-Create an AGENTS.md file with project-specific instructions for the agent.
/review-Trigger a code review of the current changes.

Sandbox

CommandDuring taskDescription
/setup-default-sandbox-Configure and install the elevated agent sandbox.
/sandbox-add-read-dir-(Windows) Grant the sandbox read access to an additional directory.

Appearance and UX

CommandDuring taskDescription
/theme-Choose a syntax highlighting theme.
/vim-Toggle Vim keybindings in the composer.
/keymap-Remap TUI keyboard shortcuts.
/pets (alias /pet)-Choose a terminal pet character or hide it.
/titleyesConfigure the terminal title bar items.
/statuslineyesConfigure the status line items.
/ideyesInclude IDE context (open files, current selection) from a connected editor plugin.

System

CommandDuring taskDescription
/statusyesShow session config (model, provider, context usage, token counts).
/appyesContinue the current session in the ByteAsk desktop app (macOS/Windows).
/psyesList background terminal processes started by the agent.
/stop (alias /clean)yesStop all background terminal processes.
/debug-configyesShow the effective config across all layers (defaults, user config, profile, CLI overrides).
/feedbackyesPackage session logs and send them to the ByteAsk team.
/quityesExit ByteAsk.
/exityesAlias for /quit.

Notes

  • Custom slash commands are not supported. The set above is fixed in the binary.
  • Commands that must exit the TUI (/login, /logout, /archive, /delete) write a marker file to ~/.byteask/ and exit; the byteask wrapper reads it and performs the action (re-auth, relaunch, etc.) before returning control.
  • /usage behaves differently on a ChatGPT-backend account: there it accepts /usage [daily|weekly|cumulative] to show token activity instead of the ByteAsk plan card.