kb
Knowledge base — articles, search, ingest, crawl.
rogeriq kb
bashrogeriq kb list [--limit 50] [--cursor <cursor>]rogeriq kb get kba_xxx
bashrogeriq kb create --title "How to refund" \ --content "Customers can request..." # inlinerogeriq kb create --title "Onboarding" --content-file ./onboarding.md
Requires write (or kb:write) scope. Pass --status published to
publish immediately and index for AI agent retrieval.
bashrogeriq kb update kba_xxx --title "New title"rogeriq kb update kba_xxx --content-file ./updated.mdrogeriq kb update kba_xxx --status archived
bashrogeriq kb publish kba_xxx
Sets status to published, sets published_at, and re-vectorizes the
article for the agent's retrieval system.
bashrogeriq kb search "refund policy"
Hybrid semantic + keyword search. Top 10 most relevant articles.
bashrogeriq kb ingest https://example.com/help/refundsrogeriq kb ingest ./local-article.md
URL → fetches the page and creates an article from the cleaned content. File path → creates an article with the file as the body.
bashrogeriq kb crawl https://docs.example.comrogeriq kb crawl list # all jobsrogeriq kb crawl status crawl_xxx # one job
Starts an async crawl job that ingests every reachable page on the
same hostname. Monitor via crawl status. Cancel via the dashboard.
bashrogeriq kb categories
bashrogeriq kb delete kba_xxx
Destructive. Removes the article and its vector entries.
Common workflows
Migrate docs from a static site
bashrogeriq kb crawl https://help.acme.com# Wait for completionwhile [ "$(rogeriq kb crawl status crawl_xxx --json | jq -r '.status')" = "running" ]; do sleep 10done
Bulk-import markdown files
bashfor f in ./docs/*.md; do rogeriq kb ingest "$f"done