@usedocent/cli

Uploads your app's routes and your Markdown documentation to Docent, so the guide can navigate your app and answer from your own docs.

npx @usedocent/cli sync ./src/app --docs ./docs --key dk_sec_...

• ./src/app is your Next.js App Router directory. A project root works too; the tool looks in app and src/app. Other routers are not read yet.
• --docs is a directory of Markdown files (.md, .mdx, .markdown). Markdown only: pages written as React components are not read; export them to Markdown first. Without --docs only routes are uploaded.
• --key is the secret key of one environment, from Settings, Environments in the Console. It can also come from DOCENT_SECRET_KEY. It is never read from a file.

The tool prints every route before it sends anything; --dry-run stops there. After an upload, indexing starts on its own and the Console's Knowledge page shows progress as the retrievable count catches up with the chunk count. The project map at https://app.usedocent.com/map draws the routes.

crawl

For a docs site with no Markdown export, crawl asks Docent's own servers to read it into knowledge.

npx @usedocent/cli crawl https://docs.example.com --key dk_sec_...

The crawl runs on Docent's servers, not on your machine: same origin as the URL you pass, over https, honouring robots.txt, up to --max-pages pages (default 200, max 200). Pages that need JavaScript to render are not read; the job reports that rather than uploading nothing silently.

• --max-pages <n> caps how many pages the job reads, 1 to 200.
• --dry-run reads robots.txt and sitemap.xml itself and prints the pages it would send and the pages robots.txt blocks, then stops. Nothing is crawled.
• --key and --endpoint work the same as for sync.

Without --dry-run, the command prints the crawl id and a link to the Console's Knowledge page, where pages and retrievable chunks appear as the job runs.