Commands

Run lucitra with no arguments to open the shell, or lucitra <command> to run one thing and exit.

Reading

Command	What it does
`plan`	Scope a goal into tasks without running any of them
`chat`	Ask Lucitra something, in this repository's conversation
`ask`	Search the project brain, falling back to the model when it is silent
`survey`	What Lucitra can work out about this repository without being told
`candidates`	Work this repository already says needs doing: issues, drafts, old TODOs
`tasks`	What has run and what is running, newest first
`dashboard`	Where the work stands across every repository, and what the week cost
`search`	Find a line across every conversation in this repository
`models`	Which model each agent runs, and on which runtime
`mcp`	Which MCP servers each agent may reach
`agents`	The team, and what each agent is configured to do

These change something, so they are the ones to read twice.

Command	What it does
`seed`	Write what the survey found into this repository's brain, as one page
`start-build`	Start a task in its own worktree. Returns as soon as it has an id
`approve`	Let a waiting task push. Omit the task when only one is waiting
`decline`	Refuse a waiting push. The worktree is left alone
`configure`	Change one setting on one agent: `configure <agent> <setting> <value>`
`reset`	Put one setting back to its default: `reset <agent> <setting>`

Command	What it does
`lucitra mcp`	Serve these commands to an agent over MCP
`lucitra browser install`	Let the Lucitra Companion extension reach this agent
`lucitra stop`	Stop the daemon. The next run starts a fresh one