article-exporter

Use the actionbook CLI fetch command with the article URL, then let the AI reformat the plain text into structured Markdown. The workflow downloads images locally, rewrites references to relative paths, and creates an index.md navigation file.

Fetch the tweet URL with actionbook using the --wait-hint heavy flag, then apply AI reformatting to reconstruct headings, code blocks, and lists. Twitter's non-semantic HTML loses structure during extraction, so the AI reformatting step is mandatory.

Dynamic sites like Medium and Twitter render content asynchronously, so extraction happens before the page finishes loading. Use the --wait-hint heavy flag and verify your actionbook CLI version is 0.9.1 or higher.

Zero-byte images usually mean the URL expired or requires a different format parameter. Retry with ?format=jpg&name=orig appended for Twitter images, or add a Referer header if the server returns 403.

Yes, but you must add 3-5 second random delays between requests to avoid being flagged as a bot. Keep batches to 5-10 articles and always test with a single article first.

The exporter cannot bypass paywalls; it only works if you are already logged in through your browser session. Content behind authentication you lack access to will export as truncated or limited text.