The Lytos CLI

The CLI is how Lytos lands in a real repository. One lyt init scaffolds the five pillars as folders, detects your stack, writes the right adapter file for your AI tool (CLAUDE.md, .cursorrules, AGENTS.md…), and installs a pre-commit hook so branch naming stays consistent.

Install

npm install -g lytos-cli

Or without installing:

npx lytos-cli init

The lyt binary is the short name (also lytos and lytos-cli if you prefer).

Commands

Command	What it does
lyt init	Scaffold .lytos/ — interactive, detects your stack, offers EN or FR
lyt board	Regenerate BOARD.md from the issue frontmatter
lyt lint	Validate the .lytos/ structure and content
lyt doctor	Full diagnostic — missing files, broken links, stale memory, health score
lyt show ISS-XXXX	Display issue detail with progress bar
lyt start ISS-XXXX	Start an issue — move to in-progress, create branch, update board
lyt claim ISS-XXXX	Assign an issue to yourself (also checks origin freshness to avoid concurrent claims)
lyt close ISS-XXXX	Close a validated issue — promote it to 5-done from 4-review (or explicitly from 3-in-progress)
lyt close	Batch-close every issue in 4-review/ — asks to confirm, --yes skips
lyt upgrade	Pull the latest bundled skills and rules into .lytos/

Key references

Overview	What the CLI does, at a glance
lyt init	Everything the scaffolder does — adapter files, stack detection, pre-commit hook
lyt board	How the kanban view is regenerated from issue frontmatter

Design principles

• Offline-first — every command except lyt init (fetches bundled files) and lyt claim/start (checks origin) works without network.
• Output for humans AND machines — every command has a human default and a --json flag for CI and scripting.
• Fail with context — errors say what’s wrong, where, and how to fix it.
• Single responsibility per command — lint validates, doctor diagnoses, they don’t overlap.