Oko/Docs

Command reference

oko requirements find

The N messages of his and the agents' nearest in meaning to a question, from the meaning index Oko keeps itself, each with its session, message index, the passage that matched and the window it lies in.

Invocation#

Command
oko requirements find QUESTION --top N [--json]

Inputs and options#

  • QUESTION is free text in any language the embedding model reads, read the way he types. --top N is how many messages are answered; a message answers once, by its nearest passage.

Output and state effect#

  • Embeds the question with the same model, scores every kept passage and prints hits ({score, session, transcript, index, speaker, text: the passage verbatim, window: {first, last, summary, digest}}) and everyWindowChars, what reading every window would show. No model is asked.

Refusals and safe exits#

  • `the embedding model <name> is not installed in <folder>: run `oko requirements embedder install`` and `the question <question> has no word the embedding model <name> knows` exit 1. A missing question is `oko requirements find requires the question, quoted as one word: oko requirements find "QUESTION" --top N`, a second word `oko requirements find acts on one <id>; it was also given <word>, which it would not read`, and a missing --top `oko requirements find requires --top <whole number>` (each exit 2).

CLI reference#