Skip to content
fb

Guides

One guide per part of fb: sessions and tiers, pages, posts, media, groups and events, the claim graph, stores, captures, and the HTTP and MCP surfaces.

Start with sessions and tiers. It is the one that shapes every other read: what tier 0 gives you, what the two cookies add, and what Facebook does when it decides you have asked enough.

  • Sessions and tiers: what reads without anything, what needs two cookies, and how Facebook says no.
  • Pages and profiles: one read for the whole shell, the About tab, and the timeline's one post.
  • Posts and comments: a permalink whole, with entities, comments and all seven reaction types.
  • Media: photos with alt text and album neighbours, videos with captions and transcripts, and the bytes.
  • Groups and events: the group shell and its discussion, and events from the tab or a permalink.
  • Finding your way in: classify anything with fb id, ask fb explain what fb would do, and find seeds.
  • The claim graph: what a read said about everything else, walked with fb graph and written as RDF.
  • Building a store: fb crawl into SQLite, fb query in plain SQL, fb export as RDF.
  • Capturing a read: the HTML, the Relay payloads and the headers, which is how a bug gets reported.
  • HTTP and MCP: the same eighteen operations over a port or over stdio.
Sessions and tiers Tier 0 needs nothing at all. Tier 1 is two cookies you paste in yourself, and this is what they buy. Pages and profiles Read a page or profile whole, read its About tab, and walk its timeline as far as the tier allows. Posts and comments Read a post whole, with its message entities, its comments, and its reactions broken out by type. Media Photos with their alt text and album neighbours, videos with captions and transcripts, and the bytes with a provenance sidecar. Groups and events The public shell of a group and its discussion, and events from the tab or from a permalink. Finding your way in Classify anything you paste with fb id, ask what fb would do with fb explain, and find seeds with fb discover. The claim graph Every read makes claims about other things. fb edges prints them, fb graph walks them, and fb rdf writes them in a vocabulary something else can read. Building a store fb crawl walks the claim graph into SQLite, fb db stats says what landed, fb query is plain SQL, and fb export writes the whole thing as RDF. Capturing a read fb archive writes one read to disk whole: the HTML, every Relay payload, the request headers, and the record fb parsed out. HTTP and MCP The same eighteen read operations over HTTP as NDJSON, or over stdio as MCP tools, generated from the operations rather than written twice.