Slash commands
Type / in a session to open the command menu; it filters as you type, and Tab completes. Arguments in <angle brackets> are required, [square brackets] optional, and a|b means pick one. Commands marked prompt become a carefully worded request to the agent instead of acting locally.
The menu also lists your own skills and custom commands and prompts from MCP servers (/mcp__<server>__<prompt>).
Help and status
| Command | What it does |
|---|---|
/help (also /h, /?) | Scrollable help with every command and key. |
/powerup [number] | Short lessons on features that are easy to miss. |
/keybindings [edit] | The keys in force; edit creates your keybindings.json and shows its path. |
/status | Account, plan, session, model, mode and MCP status. |
/doctor (also /checkup) | A health check of the install and environment, including saved sessions that are damaged and what to do about them. |
/version | The syntic version. |
/release-notes | What is new in this version. |
/feedback [report] (also /bug) | Save a feedback report with session details on this computer. Nothing is uploaded. --with-transcript includes the conversation with likely secrets masked. |
/debug [off|<problem>] | Turn on debug logging, optionally describing what to troubleshoot. |
Usage, cost and plan
| Command | What it does |
|---|---|
/cost | This session’s spend: the total, each figure labelled reported, calculated or estimated, then tables by model and by agent (sub-agents included). |
/stats | Token counts and spend by model, with an activity view. |
/usage | Plan limits, cost, trends and plugin usage in tabs. |
/insights | What drove token use over the last day and week. |
/context | How full the context window is, and what fills it. |
/rate-limit-options | Where each plan limit stands, when it resets, and your options. |
/usage-credits (also /extra-usage) | Your organization’s usage-credit balance and whether credits are on. |
/upgrade | Your plan’s usage, and the billing page. |
The conversation
| Command | What it does |
|---|---|
/clear [name] (also /new, /reset) | Start a new conversation. The old one stays resumable (and gets name, if given). |
/compact (also /condense) | Summarize older messages to free context. |
/autocompact [on|off|<size>|default] | When automatic compaction happens. |
/recap | One-paragraph catch-up on recent activity. |
/summary | prompt. A summary of the conversation so far. |
/btw <question> | prompt. A side question whose answer is not added to the main conversation. |
/rewind [n] (also /undo) | Go back to an earlier message, optionally restoring files. Same as Esc Esc. |
/checkpoint [label] | Bookmark this point; without a label, opens rewind. |
/rollback | Undo the last message turn. |
/copy [N] | Copy the latest reply (or one N back), or one of its code blocks. |
/export [md|json|text|<file>] | Export the conversation. |
/share [file] | Save a Markdown copy to share, with likely secrets masked. |
/files | Files read or changed in this conversation. |
/diff [git [path]] | What this session changed; git shows the working tree diff. |
/todos | The agent’s current to-do list. |
/goal [<condition>|clear] | Keep working across turns until a condition is met. |
/loop [interval] <prompt> · /loop list · /loop stop [id] (also /proactive) | Re-run a prompt on an interval while idle. |
/schedule [list | add … | pause <id> | resume <id> | delete <id>] (also /routines) | Scheduled routines for this project. |
/workflows [stop <id>] | Workflow runs in this session. |
/thinking | Show reasoning in full or as a preview. |
/focus | Condensed view: prompts, short tool lines and answers. Interactive only. |
Sessions
| Command | What it does |
|---|---|
/resume [text or pull request link] (also /continue) | Reopen an earlier session. |
/sessions (also /session) | Browse saved sessions. |
/rename [title] | Rename this session; with no title, a name is written for you. |
/branch [name] (also /fork) | Continue in a copy of this conversation. --bg <task> runs the copy in the background. |
/background <task> (also /bg) | Hand a task to a background session that starts from this conversation. |
/tasks (also /bashes) | The agents view: subagents, background runs and other running work. |
/cd <folder> | Move the session to another folder (asks to trust it first if needed). |
/exit (also /quit, /q) | Leave. /exit remove also removes a --worktree session’s worktree. |
Model and behavior
| Command | What it does |
|---|---|
/model [name] | Pick the model (--session to keep it to this session). The list comes from the Syntic platform; models your plan does not include are shown with the plan they need. |
/fast | Fast mode on or off. |
/effort [level] | Reasoning effort: low, normal, high, xhigh, max; no level cycles. ultracode on|off is a separate switch. |
/plan [task] | Plan mode on or off; with a task, switch and send it. |
/output-style [style] | How replies are written. |
/advisor [on|off|default|<model>] | The experimental Advisor tool for this session. |
Look and feel
| Command | What it does |
|---|---|
/theme [create <name> [dark|light]] | Pick a theme with live preview, or create your own. |
/color [name|default] | A color for this session’s prompt bar. |
/tui [inline|fullscreen] | Inline or full-screen view, from the next start. |
/scroll-speed [0.25-10|default] | How far one mouse-wheel notch scrolls. |
/vim | Vim keys in the prompt on or off. |
/statusline [instructions] | prompt. Set up a status line for you. |
/terminal-setup | Make Shift+Enter insert a new line in this terminal. |
/settings (also /config) | The settings panel. /config key=value sets one value directly. |
Project and context
| Command | What it does |
|---|---|
/init [focus] | prompt. Study the project and write a SYNTIC.md guide. |
/memory | Memory files in this session, and the auto memory switches. |
/remember <fact> | Save a fact to memory (starting a message with # does the same). |
/add-dir <path> [--remember] | Let this session work in another folder. |
/permissions (also /allowed-tools) · /permissions auto <allow|soft_deny|hard_deny> <rule> [--project] | Manage rules and auto mode’s rules. |
/sandbox [on|off | network on|off] | Show or switch the shell sandbox. |
/env [NAME=value | unset NAME] | Environment variables for this session’s shell commands. |
/ide [open] | List editors and connect to one. |
/skill-doctor | Which skills cost context and which went unused. |
Agents, skills, plugins, hooks and MCP
| Command | What it does |
|---|---|
/agents [status] | Running agents, and your custom agent library (create, edit, delete, use). |
/skills | Browse available skills. |
/reload-skills | Re-scan skills and custom agents without restarting. Interactive only. |
/plugin (also /plugins) | Plugins and marketplaces: installed, discover, marketplaces. |
/reload-plugins | Re-scan plugins, skills and agents. |
/hooks · /hooks add [--project|--local] <Event> [matcher] -- <command> · /hooks remove <n> | Browse, add and remove hooks (Hooks reference). |
/mcp [reconnect all] | MCP servers, their tools and sign-in. |
Everyday helpers
All of these are prompt commands: each sends a carefully worded request, and the work then shows up in the conversation like any other turn.
| Command | What it does |
|---|---|
/subtask <task> | Hand a side task to a background sub-agent that starts with a copy of this conversation, and keep going here. You get a one-line note when it starts and a short result when it reports back. |
/verify [what] | Check that the latest change really works by building and running the software (not only reading code or running tests), then report the commands, what was seen and a verdict. Nothing is edited. |
/run [what] | Start the project’s app and show the change working. |
/update-config <request> | Describe a settings change in plain words (“allow cargo test”, “turn off auto-compact”); the right settings file is edited. |
/fewer-permission-prompts [focus] | Look through this project’s past sessions for read-only commands and MCP tools you keep approving, show them ranked with a count, then add narrow allow rules for them to .syntic/settings.json. |
/auto-mode-setup [focus] | Draft auto mode rules (allow, soft_deny, hard_deny) that fit this project, each with a reason, and save them to your user settings only once you confirm. Interactive only. |
/import [<source>] [--dry-run] [--yes] | Bring instruction files, MCP servers and saved prompts or commands over from another coding assistant’s configuration into SYNTIC.md, .mcp.json and .syntic/commands/. Lists what it found and asks before writing; --dry-run only lists, --yes imports everything. Existing Syntic files are merged, never overwritten, and secret values are not copied. |
Code work
All of these are prompt commands.
| Command | What it does |
|---|---|
/review [PR number or focus] | Review the current changes or a pull request. |
/code-review [low|medium|high|max] [--fix] [--comment] [--max-findings N|all] [target] | A bug hunt in a background subagent. |
/ultrareview [--fix] [target] | Three review passes (correctness, security, tests) merged into one list. |
/security-review | Look at the branch’s pending changes for security problems. |
/simplify [focus] | Clean up changed code and apply the fixes. |
/commit [instructions] | Commit the current changes. |
/commit-push-pr [instructions] | Commit, push and open a pull request. |
/pr-comments | Fetch the review comments on the current pull request. |
/autofix-pr [number|url] | Fix a pull request’s failing checks and review comments locally. |
/batch <task> | Split a task across parallel agents, each in its own worktree. |
/ultraplan <task> | A long-form plan written to your plans folder by a background subagent. |
/init-verifiers | Set up checks the agent can run to verify its work here. |
/team-onboarding [role or focus] | Write an onboarding guide for a new teammate. |
/deep-research <question> | Research in steps and answer with sources. |
Account
| Command | What it does |
|---|---|
/login | Sign in to Syntic (/login copy copies the sign-in address). |
/logout | Sign out. |
In -p runs, commands that need the interactive screen (such as /compact or /rename) are refused with a short message, and an unknown /command exits with an error that suggests running syntic interactively and typing /help.