Type / in a session to list commands under the input; type to filter. /help lists commands and keys. Your own commands (.neuralinverse/commands/*.md), plugin commands (/<plugin>:<name>) and MCP prompts (/<server>:<prompt>) appear alongside the built-in ones; organisation commands delivered by a signed policy show org as their source.
Conversation
| Command | What it does |
|---|---|
/new | Begin a fresh conversation; settings stay |
/condense | Replace the history with a summary to make room |
/copy | Put the latest reply on the clipboard |
/undo | Take back the last message and the file changes it caused |
/restore | Return to the point before an earlier message |
/remember <note> | Save a standing instruction to AGENTS.md |
/attach-image | Send the clipboard image with your next message |
/plan | The steps the agent is working through |
/rename | Give this conversation a title (shown in /history) |
/aside | Ask a quick side question without tools; the conversation is left untouched |
/summary | A short summary of the session so far; the conversation is left untouched |
/fork [title] | Copy this conversation into a new session and continue there; the original stays as it is |
/deep-plan <task> | A planning-only turn in read-only mode with the plan role model. It may ask you questions, saves a structured plan to .neuralinverse/plans/<date>-<slug>.md and offers to carry it out |
/voice | Speak your next message: records until you stop talking, then sends the transcription. Needs the voice setting |
Model and mode
| Command | What it does |
|---|---|
/model | See or change the model (live list from every provider) |
/mode | See or change what the agent may do without asking |
/rules | List allow and deny rules, or add one: /rules allow <rule>, /rules deny <rule> (--project or --user to save) |
/effort | How hard the model thinks before answering |
/show-thinking | Show the model's reasoning in the transcript |
/vim | Turn vim-style keys in the prompt on or off |
/output-style | How answers are written: default, concise, explanatory, review or your own |
/provider [add|switch|test|remove] | Manage model providers. See Providers and models |
Files and history
| Command | What it does |
|---|---|
/history | Saved conversations in this folder (--tag <name> shows only tagged ones) |
/open | Continue a saved conversation |
/export | Write this conversation to a Markdown file |
/instructions | Instruction files in use; opens AGENTS.md in your editor |
/bootstrap | Have the agent study the repository and write AGENTS.md |
/review | Get a review of your uncommitted changes |
/add-dir | Let tools use another folder for this session (--save keeps it for the project) |
/diff | Show uncommitted changes (staged and unstaged), optionally for one path |
/files | Files the agent changed in this session, with line counts |
/commit | Have the agent write a commit message and commit your changes. Never pushes. Runs the checks commit gate first |
/pr | Have the agent push this branch and open a pull request with gh |
/tag <name>, /tag -<name> | Tag or untag this conversation |
/share [file] | Export this conversation as one self-contained HTML page with secrets redacted |
Tools and integrations
| Command | What it does |
|---|---|
/tools | What the agent can use |
/mcp | MCP servers: status, logs, reconnect <name>, sign in or out |
/skills | Skills the agent can load |
/agents | Agent profiles the agent can delegate to; new <name> creates one |
/hooks | Hooks that run on agent events, and where they are set |
/tasks | Background helpers and shell jobs; stop <id> ends one |
/index | Code index status; rebuild scans the workspace again |
/security-review | Security review of your changes (read-only), findings ranked by severity |
/reload-plugins | Re-read installed plugins without restarting; MCP servers reconnect |
/environment | Choose this project's developer environment: firmware (embedded), hardware (hdl) or none |
/checks | Findings of the checks gate runs in this session (/checks all adds coverage) |
/pr-comments [number] [--all] | Attach a pull request's unresolved review comments to your next message |
/autofix-pr [number] | Have the agent fix unresolved review comments and failing checks on the current branch. Never pushes |
/github-setup | Add a workflow that answers @neuralinverse in issues and pull requests, and set the key secret with gh (asks first) |
/issue [title] | Draft a GitHub issue with the agent and create it after you confirm |
The GitHub commands are described in GitHub.
System
| Command | What it does |
|---|---|
/help | List commands and keys |
/info | Session, model and connection overview |
/usage | Tokens used and estimated spend so far |
/window | How full the model's context window is, and the split of the next request. (The older /context name now belongs to organisation context.) |
/stats | Session statistics: time, turns, tools, tokens, spend and lines changed |
/config | Settings in effect and where they came from, secrets hidden |
/plugins | Active plugins, plus any load problems |
/doctor | Check the installation, keys, tools and connections |
/sandbox | Run shell commands in an operating-system sandbox for this session |
/theme | Switch the colour theme (saved in your user settings) |
/keys | Keyboard shortcuts of the layout in use |
/keybindings [show|reset] | Customise key bindings in ~/.neuralinverse/keybindings.json |
/terminal-setup | Make Shift+Enter insert a new line (and Option work as Alt on macOS) in VS Code, Cursor, Windsurf, kitty, Ghostty and Terminal.app, after showing the change and asking; backups are made. iTerm2, WezTerm, Windows Terminal and Alacritty get manual steps |
/env | What the agent's environment holds: provider variables (only whether they are set), proxy settings (credentials masked), sandbox, shell, and tools on PATH |
/statusline [template|command|off] | Set a custom status line: built-in scripts written to ~/.neuralinverse/statusline.sh, or your own command |
/update | Check for a newer version and install it (asks first) |
/feedback | Write a bug report or feedback file to attach to an issue. Nothing is sent |
/release-notes | Release notes |
/version | Print the version |
/quit | End the session |
/login, /logout | Sign in to or out of Govern |
/account | Your account, Account ID, region, server, session, policy and connection state, console pages |
/policy [refresh] | Your organisation's signed policy, or fetch it now |
/keystone-context <args> | The neuralinverse context commands inside a session. See Organisation context |
/activity [log [n]|flush] | What activity capture records about this session, and why it is on or off |