Command line
e help prints this list, with any flags and commands your extensions add.
On this page
Start a session
| Command | What it does |
|---|---|
e [message] | start a session in this directory, optionally with a first prompt |
e -c | continue this directory's most recent session |
e -r | pick a saved session to resume |
e -p [message] | run one turn headless and print the reply |
e rpc | run the headless session server over stdin and stdout |
A plain e does not read piped stdin. Use e -p for one turn, or e rpc for
a client of your own. See automation.
Run options
| Option | What it does |
|---|---|
-p, --print | run one headless turn. The prompt is the argument or piped stdin. |
-j, --json | machine output for --print, doctor, providers, and --version. With -p, one JSON line per event and a result line. |
-m, --model <model> | use this model for the process |
--ef, --effort <level> | set reasoning effort for the process |
-i, --image <path> | attach an image to the first prompt, repeatable |
-P, --package <source> | load a package for this run only, repeatable |
--ne, --no-extensions | start without extensions |
--nt, --no-tools | expose and run no tools |
--ns, --no-save | keep the conversation in memory only |
Commands
| Command | What it does |
|---|---|
e docs [topic] | print a built-in guide |
e update | update to the latest release |
e install [source] | install a package, or make every listed one current |
e remove <source> | forget a package and delete its clone |
e packages | list installed packages |
e packages init <dir> | start a package to publish |
e trust [dir] | trust a workspace's AGENTS.md, skills, and prompts |
e untrust [dir] | stop loading them for that workspace |
e auth | show sign-in status |
e providers | list provider support and sign-in state |
e doctor [--no-network] | print diagnostics that are safe to paste |
e help | print the help |
e -v, e --version | print the version |
Packages has its own guide. For e trust, see
instructions.
Inside a session
| Command | What it does |
|---|---|
/login | sign in to a provider with an account or API key |
/models | switch the model |
/effort | show or set reasoning effort |
/scoped-models | choose which models ctrl+p cycles |
/reload | reload extensions, themes, and config |
/resume | resume a saved session |
/tree | rewind to an earlier message and branch in place |
/new | start a fresh session |
/fork | continue in a new session file seeded with this one |
/export | write this session as one self-contained HTML page |
/copy | copy the last reply |
/compact | summarize into a fresh session. /compact <focus> steers what it keeps. |
/usage | tokens and estimated cost by model. /usage 24h narrows the window. |
/undo | put back what the last write or edit replaced |
/trust | trust this directory |
/settings | change preferences |
/help | show these commands |
/quit | exit |
The list follows what you install. Each prompt template becomes a /name
command of its own. See prompt templates.
An extension adds commands the same way.