# Mnemo: commands and shortcuts

Mnemo 0.1.9. Type `/` in the prompt for the command menu, or `/help` for the same list in the app.
The current version of this list is at https://mnemo.atmanmishra.dev/shortcuts.

## Slash commands

| Command | What it does |
|---|---|
| `/help` | commands and keys |
| `/model` | choose the model |
| `/thinking` | set how hard the model thinks |
| `/login` | add a provider (API key or subscription) |
| `/logout` | remove a provider's credentials |
| `/new` | start a fresh session |
| `/resume` | continue an earlier session in this folder (type to search what was said in it) |
| `/rewind` | go back to before an earlier message: the conversation, the files, or both (also `esc esc`) |
| `/compact` | summarise the conversation to free context |
| `/cost` | tokens and spend for this session |
| `/stats` | this session, the last 7 days, and how the tools fared |
| `/clear` | clear the screen (the session is kept) |
| `/memory [words]` | what Mnemo remembers; with words, search it |
| `/remember <thing>` | tell Mnemo something to keep, e.g. `/remember we deploy with fly` |
| `/forget <key>` | retire a remembered fact, e.g. `/forget package manager` |
| `/mode` | how much Mnemo may do without asking: default, accept-edits, plan, yolo |
| `/yolo` | toggle bypass: ask for nothing (deny rules in permissions.json still hold) |
| `/accept-edits` | toggle accept edits: change project files without asking; commands still ask |
| `/plan` | toggle plan mode (read-only) |
| `/theme` | switch the look: Mnemo Night, Dracula, Nord, Gruvbox, Tokyo Night, Catppuccin, Synthwave, Amber CRT, Game Boy, Paper, Solarized |
| `/settings [name value]` | change how Mnemo looks and behaves: theme, keys, agent limits, notifications; `/settings hops 20` sets one, `/settings reset` undoes them |
| `/mouse` | mouse wheel scrolls the conversation (on), or your terminal selects text (off) |
| `/bestof <n> "<check>" <task>` | race n attempts at a task and keep the smallest change that passes the check |
| `/skills` | the skills Mnemo can load |
| `/reload` | reload skills, prompts and extensions |
| `/update` | install the latest Mnemo; `/update auto on\|off` turns the daily check on or off |
| `/quit` | leave Mnemo |
| `/skill:<name>` | run a skill |
| `/<template>` | run a prompt template |

## Keys in the prompt

Fourteen of the chords on this page can be changed in `/settings` (the Keys group): the palette, sidebar, find, expand, clear, thinking level, hub, playground, project switch, new agent, split, next layout and previous and next agent. The defaults are shown here; `/help` and the `?` panel show the ones you set.

| Key | What it does |
|---|---|
| `enter` | send; while Mnemo works, the message is queued |
| `alt+enter` | new line (or end a line with `\`) |
| `esc` | interrupt the turn; close a menu; twice clears the draft |
| `esc esc` | on an empty prompt: rewind (see `/rewind`) |
| `↑` `↓` | history (or move in a menu) |
| `tab` | accept the highlighted suggestion |
| `@` | mention a file |
| `/` | commands |
| `shift+tab` | cycle mode: default → accept edits → plan |
| `ctrl+t` | cycle the thinking level |
| `ctrl+o` | expand thinking and tool output, including what is already printed |
| `ctrl+f` | find in the conversation (type; `↑` `↓` step through matches; `enter` stays there; `esc` closes) |
| `ctrl+k` | on an empty prompt: go to a command, agent, session or file |
| `ctrl+l` | clear the screen |
| `ctrl+c` | clear the draft; interrupt while working; twice to quit |
| `ctrl+d` | quit when the prompt is empty |
| `?` | shortcut panel (on an empty prompt) |
| `←` `→` `home` `end` `ctrl+a` `ctrl+e` | move the cursor |
| `ctrl+u` `ctrl+k` `ctrl+w` `alt+backspace` | delete to start, to end, or a word |
| `alt+b` `alt+f` | move by word |

## Workspace (full screen)

| Key | What it does |
|---|---|
| `ctrl+b` | show or hide the sidebar |
| `tab` | move focus: prompt → sidebar → transcript |
| `esc` | back to the prompt |
| `PgUp` `PgDn` | scroll the transcript from anywhere |
| `[` `]` (or `←` `→`) | step through turns on the timeline (when the transcript has focus) |
| `1`–`5` (sidebar) | switch pane: files, memory, sessions, skills, logs |
| `↑` `↓` (sidebar) | move the selection; the main area previews it |
| `→` `enter` (sidebar) | open a folder or preview; `←` closes it |
| `i` (files pane) | put the file in the prompt as `@path` |
| `r` (sessions pane) | resume that session |
| `m` (memory pane) | open the memory map; arrows move between nodes, `enter` opens one |
| `e` (memory pane) | edit a fact or a pitfall's fix |
| `d` (memory pane) | forget that item |
| `+` `-` (memory pane) | say an item helped, or was wrong |
| `x` (preview) | close the preview |

## Agents (several projects at once)

| Key | What it does |
|---|---|
| `ctrl+g` | hub: a card per agent (again to go back) |
| `ctrl+p` | switch project, or open one by typing its path |
| `ctrl+n` | another agent on the same project, in parallel |
| `ctrl+s` | split view (never more than four panes); again for single view |
| `alt+l` | in a split: the next layout (columns, rows, grid, main + side) |
| `ctrl+y` | playground: every agent, sub-agent and message between them, to watch only |
| `c` `tab` (playground) | the selected agent's conversation with another as a thread · its next partner |
| `alt+1` … `alt+9` | go to agent 1–9 |
| `alt+,` `alt+.` | previous / next agent; in a split, every pane, so you can type to a running sub-agent |
| `esc` (sub-agent pane) | give the keys back to the agent |

In the hub: `←` `→` `↑` `↓` choose a card, `enter` opens it, `n` new agent, `s` split with it, `x` close it, `l` then `l` on another card links the two as siblings (they can see and message each other), `u` unlinks, `f` splits with its family, `p` opens the playground. Each card shows its sub-agents and links.

## Dialogs

| Key | What it does |
|---|---|
| `↑` `↓` `tab` | move through choices; type to filter |
| `enter` | choose |
| `esc` | cancel |
| `y` / `n` | answer a yes-or-no question |

## Command line

| Command | What it does |
|---|---|
| `mnemo` | start in this folder |
| `mnemo -c` / `--continue` | continue the most recent session here |
| `mnemo -p "<prompt>"` | answer once and print the result (no interface) |
| `mnemo doctor [--fix]` | what is installed, configured and reachable; `--fix` repairs what it can |
| `mnemo install [--yes]` | put this copy and its `memsrv` in `~/.mnemo/bin` and on your PATH (a downloaded `mnemo.exe` offers this on its own) |
| `mnemo update [--check] [--auto on\|off]` | install the latest release; `--check` only says if one exists; `--auto` sets the daily check (off by default) |
| `mnemo settings [set <name> <value> \| reset [name]]` | the same settings as `/settings`, from a shell; with no arguments, lists them and what differs from the default |
| `mnemo uninstall [--deep] [--yes]` | remove the program; `--deep` also deletes memory, sessions, credentials and settings |
| `mnemo memory …` | memory for other agents: `setup claude-code\|codex`, `ingest`, `tidy [--apply]`, `curate [--apply]`, `status` |
| `mnemo --demo` | a scripted session in a scratch project; needs no key |

Options: `--cwd <dir>`, `--plan`, `--accept-edits`, `--yolo`, `--trust-project`, `--no-memory`, `--no-reflect`, `--no-verify`, `--best-of <n> --check <cmd>`, `--escalate <provider/id>`, `--model <provider/id>`, `--dump`, `--no-motion`, `--no-boot`, `--inline`, `--intro`, `-v`, `-h`.

## Environment variables

| Variable | What it does |
|---|---|
| `MNEMO_HOME` | where Mnemo keeps its files (default `~/.mnemo`) |
| `MNEMO_PASS_ENV=NAME,…` | give named variables (an API key, `GITHUB_TOKEN`) to commands the agent runs; they get none otherwise |
| `MNEMO_KERNEL_MEMORY_MB` | Python kernel memory cap on Linux (default 8192; 0 = none) |
| `MNEMO_MOUSE=0` | leave the mouse to the terminal |
| `MNEMO_LINKS=0` | no clickable links |
| `MNEMO_NOTIFY=0` | no desktop notifications when a turn ends in the background |
| `MNEMO_EMBED_URL`, `MNEMO_EMBED_MODEL`, `MNEMO_EMBED_KEY`, `MNEMO_EMBED_TIMEOUT_MS` | use your own embedding server (e.g. Ollama) for memory search |
| `MNEMO_MEMORY_MODEL`, `MNEMO_ESCALATE_MODEL` | models for memory reflection and escalation |
| `MNEMO_PYTHON` | which Python runs the `ipy_run` tool |