The VibeCop CLI ships as the npm package @maiife-ai-pub/vibecop, installable globally or via npx, and exposes a single vibecop binary. It runs the same detector logic used in PR reviews locally, against any directory, with no repository connection required for a scan.
vibecop scan do?vibecop scan [path] scans a directory for AI-introduced code antipatterns and prints results in terminal, JSON, or agent-readable format. A --severity flag sets the minimum severity to report (P1-P4), and --exit-zero keeps the exit code 0 even when critical findings are present — useful in CI pipelines that shouldn't hard-fail.
vibecop login authenticate the CLI?vibecop login opens your browser to vibecop.maiife.ai/cli-auth and runs a PKCE authorization flow, so the CLI never handles your password directly. Once signed in, vibecop link connects the local scan to a specific VibeCop project, and vibecop whoami confirms which account and org the CLI is currently using.
| Command | Purpose |
|---|---|
| scan [path] | Scan a directory for AI-introduced code antipatterns |
| init | Scaffold VibeCop config for a project |
| login | Authenticate via browser PKCE flow |
| logout | Clear the local session |
| whoami | Show the signed-in account and org |
| link | Connect the local scan to a VibeCop project |
| stats | Show recent scan stats |
| serve | Start the MCP server (stdio transport) |
Yes. The CLI package bundles a GitHub Action (built via its build:action script into a standalone esbuild bundle) so a repository can run VibeCop scans as a CI step, independent of the hosted webhook-triggered PR reviews. It shares the same scanning core as the local CLI.