FAQ

Frequently asked questions

Short answers, each drawn from the README. For anything not covered here, see the README.

Is lea a replacement for Claude Code or Codex?

No. lea runs the real agent CLI, Claude Code by default or Codex, with your own settings, skills and plugins. It gives it a safe place to run, and lea deck to run several at once.

What is the difference between lea and lea deck?

lea is the tool: the lea command, its daemon, the container, the proxy, the agents, their messages and allowances. lea deck is its interface: the terminal lea opens, with your agents in a column and the selected one’s terminal beside it (lea web serves it in your browser). Both come in the one lea install, and lea works without the deck: lea config deck off. lea deck.

Which agents does it run?

Claude Code, Codex, and through Codex any OpenAI-compatible endpoint that serves the Responses API (Ollama, vLLM). Without the deck, lea config tool runs the command of your choice in the container. The Control Tower and lea’s plugin run on Claude Code only. Providers.

Why not just skip permissions on my host?

On your host, an agent that skips its permission prompts gets everything your user can reach: keys, tokens, your network. In lea it can go just as fast, but what it can reach is fixed by the container and the proxy.

How is it different from a devcontainer?

A devcontainer isolates a toolchain; lea isolates the agent. It adds an egress proxy that refuses private networks, keeps your host’s credentials out (the agents read only the logins lea gives them), nests its own container engine, audits risky changes on exit, and gives you lea deck for many agents.

Does my code leave my machine?

Only where the egress allowlist reaches: your agent’s model API, and the forges and registries your agents push to. lea has no server and no account, and sends no telemetry of its own. Usage metrics go nowhere unless you set LEA_OTEL_UPSTREAM to your own collector.

Where do I approve a database query?

In lox run, in a terminal of your own outside lea. Requests to the Locks (locksql, lockmq, lockfile, locklog, lockssh, lockhttp) are decided by the Lock Family’s hub, never by lea. lea answers only the agents’ own permission requests: in the deck, or without it with lea requests, lea approve and lea deny (or the agent’s own dialog). Add the Locks.

Does lea need lox?

Not to run agents. You need a recent lox for the Locks (lea doctor warns when it is older than the release lea pins): lox add sets a lock up in the projects directory, and lox run runs it and serves the decisions. lea starts no lox command; lea doctor only runs lox version.

Can an agent reach a service on my LAN?

Not by default. List the hosts it may reach in the internal-hosts file, one <host or IPv4> <ports> entry per line (LEA_INTERNAL_HOSTS is its path, by default config/internal-hosts in lea’s installation). Egress.

Can agents push and merge on their own?

When you say so. lea allow deliver, run on the host, lets every agent of a lea push its own branch, open its pull request and merge it, on the repositories it lists. lea allow shows what you allowed; lea allow off takes it back. Allowances.

Can I keep my Claude Code settings in lea?

Yes, on the host: lea config claude sets the default mode, the allow, ask and deny rules, the model and a few more for every agent, without editing a file; alone, it shows each setting and its source. A statusLine command in your ~/.claude/settings.json draws the agents’ status line too.

Does it run on Windows?

Through WSL 2: install and run lea from your WSL distribution, with Docker Desktop. Linux and macOS run it natively.

What does it cost?

lea is free and open source under the Apache License 2.0. You bring your agent’s subscription or API key.