Facebook, as structured data

Reading Facebook usually means the Graph API with its app review and tokens, or a headless browser and a pile of selectors that break every quarter. fb takes the third route.

A facebook.com page ships its own data inline. The Relay store the page renders itself from is sitting in the HTML as JSON, and that is the same data, not a scrape of the rendered text. fb reads that, parses it into typed records, and prints them as a table, JSON, CSV or just URLs.

fb page nasa                  # the profile record, whole
fb post <url>                 # a post with its comments and reactions
fb photos nasa --limit 20     # the photo tab
fb edges nasa                 # what that page claims about everything else
fb crawl nasa --depth 2       # walk those claims into a SQLite store

Nothing here needs a credential. Two cookies from a browser you are already signed into unlock a few more reads, and fb is explicit about which ones and why.

What it does

• Reads a page whole. Name, id, category, counts, bio, contact block, cover, avatar, the tabs it has, and the newest post.
• Reads a post whole. The message with links and mentions resolved to nodes, reactions broken down by type, share and comment counts, the media, and the comments the permalink ships.
• Reads media. Photos with alt text and album neighbours, videos with their captions and Facebook's own transcript, and the bytes with a provenance sidecar.
• Says what it did not see. Every record carries the tier it was read at, which surfaces answered, the URLs behind it, and what it missed.
• Turns pages into a graph. fb edges prints the claims one read makes, fb crawl walks them into a store, and fb export writes the lot as RDF over schema.org.
• Answers to something else. fb serve puts the same reads behind HTTP, and fb mcp puts them in front of an agent.

Read-only, and checked

Every request fb makes is a GET, except one form post that replays a query the page itself shipped. That is not a promise in a README. A test reads the package as source and fails the build if a second non-GET appears, if an operation name ends in Mutation, or if any registered operation is anything but a read.

Where to go next

• New here? Introduction is the mental model, and it is worth ten minutes before anything else.
• Ready to run something? Quick start.
• Installing? Installation.
• Looking for one task? The guides go command by command.
• Want every flag? The CLI reference.