Ask: the ten documents an agent would be handed, and why. Answer: the one passage it would cite. Documents, Words and Index: what the index is made of. Identifiers: which ID families it keeps whole.
answer
Where Ask shows the ten documents an agent is handed, this shows the one passagefux answer would cite — fetched from the source and re-scored when it is reachable, with its lines, its sha and whether the bytes are still current.
↑ ↓ to move, Enter to open. Every document in .fux/index/REGISTER.
Pick a document
Its X-ray — what came out of the bytes, what was indexed, how it links, and whether it comes back for its own title — is computed for that document alone, then cached.
how the analyzer reads a question
Every kept word shows its analyzed form, how many documents carry it (df) and its idf — the weight the scorer gives a match on it. Dropped words are stopwords the analyzer never indexes.
vocabulary — every term the index holds
Pick a word
See every document that carries it.
index
Open this tab to read the whole index.
identifier families — a test bench
Type a template (RF-{n}, WIKI-{X}-{n}) or, for experts, a regex. See what it matches across the corpus, how each match would be indexed, and whether Python and Node read it alike. Nothing is written: copy the line into .fux/identifiers.toml [user], then run fux ingest.