Does it need an account?

No. The app runs locally with your own model keys. An account only exists for hosted models and cloud deploy on Pro.

Where does my data live?

On your disk. Context and memory are plain-text files under ~/.letronna. Nothing leaves the machine unless you connect a hosted model.

Any OpenAI-compatible endpoint, Anthropic, local models through Ollama or llama.cpp, and the hosted pool on Pro.

Yes. Memory is written to disk as you work and loaded on the next launch. Edit or delete it like any other file.

Scheduled runs pause unless you have deployed to cloud. Nothing runs on your machine while the app is closed.

The agent runtime is Apache-2.0 on GitHub. The hosted model pool and cloud deploy are the paid parts.

Usar LetronnaLa terminal

Everything letronna takes on the command line: its commands and its options.

Esta página aún no está traducida, así que se muestra en inglés. Puedes traducirla en GitHub. GitHub ↗

letronna --help prints this list too.

Command	Does
`letronna`	open a conversation
`letronna -p "question"`	ask once, print the answer and exit; see Scripts and one-shot runs
`letronna gui`	open the window instead
`letronna serve`	answer chat apps and run scheduled jobs, with no window
`letronna setup`	name, sign in, provider, model, install
`letronna doctor`	check the setup, the disk and the gateway
`letronna cloud`	run a task on Letronna's servers (Pro); see Letronna Cloud
`letronna wallet`	the wallet the agent pays from, and its limits; see Wallet
`letronna mcp`	add, list, get or remove MCP servers; see MCP servers
`letronna mcp serve`	offer the browser, media and search tools to another agent over MCP
`letronna acp`	serve an editor over the Agent Client Protocol
`letronna --uninstall`	take it off again

Option	Does
`-c`, `--continue`	resume this folder's most recent session
`-r`, `--resume [id]`	choose a session, or open the one with this id (the id `/where` shows)
`--worktree`	work in a git worktree of its own, on a branch made from HEAD, leaving your checkout alone
`--mode <mode>`	how much runs without asking: `default`, `accept-edits`, `plan`, `auto`, `dont-ask` or `bypass`
`--yes`	the same as `--mode bypass`: run tools without asking, except risky ones
`--plan`	the same as `--mode plan`: read and search only
`--model <id>`	this session's model, leaving the default as it is
`--gateway <name>`	the gateway that model is on
`--effort <level>`	how hard it thinks in this session, or `auto`
`--memory <when>`	when this session saves to memory: `when-asked`, `ask` or `auto`
`--memory-scope <where>`	where it keeps memory: `auto` (everywhere), `project` or `chat`
`--private`	a new private chat: only its own notes load, and it saves only when asked
`--offline`	with `doctor`: leave the network alone
`-h`, `--help`	the list of commands and options
`-V`, `--version`	which version this is

The modes are explained in Asking first and modes, and memory in Memory.

letronna --mode accept-edits --worktree
letronna -r 16
letronna doctor --offline

With no mode option, the terminal starts in the mode the window starts sessions in.