Rank zsh history suggestions while typing

01 · Role in the system

What Jev does here

One request per keystroke carries the typed buffer and ID-tagged history candidates: a Choice asks which entry the user is completing and a Noul asks whether any candidate completes it at all. Literal prefix matches short-circuit without a request, and fuzzy mode suggests only when the top score and the Noul clear configurable thresholds, because the two signals fail in different places. Stale responses are discarded when the buffer changes mid-request, and the prompt never blocks.

02 · Control boundary

Where Jev sits

Typed buffer plus deduplicated history to one Choice and one Noul; deterministic gating, highlighting, and accept-key handling stay in zsh and Node.

Code owns the loop, permissions, thresholds, validation, and side effects. Jev owns only the bounded judgments described above.

03 · Known limits

What this evidence does not prove

• Suggestions depend on the local history file, and the end-to-end test suite needs a live API key.
• No accuracy study against real typing sessions; the demo animation uses a fabricated history.

04 · Attribution

Public sources

This is a Community record: the project was published by a third-party community author.

• mrnugget ↗Community · github · public · checked 2026-09-18