Stop the docs claiming the LLM router is off #49

2026-07-31T20:37:31+02:00

kami commented

2026-07-31 20:37:31 +02:00

Doc drift found while sweeping the Vikunja board after #47/#48.

CLAUDE.md routing section said the LLM router is wired nil and the classifier cascade is the committed default. False since #47 - voice.go:214 wires pickLLMRouter, DefaultLLMRouter is on, deploy/mavend.json sets llm_router true. Rewritten, with the measured numbers for both engines and the note that the classifier is the failure floor and must not be deleted.
CLAUDE.md persona line pointed at a memory file that does not exist, so the rule was nowhere in the repo. Written out, with the three eval checks that enforce it.
MODEL-BAKEOFF: eval-models target does exist now; the routing numbers are the production path, not a bench artifact; the truncated gguf is deleted. Struck through rather than deleted.

Docs only, no code touched.

Doc drift found while sweeping the Vikunja board after #47/#48. - CLAUDE.md routing section said the LLM router is wired nil and the classifier cascade is the committed default. False since #47 - voice.go:214 wires pickLLMRouter, DefaultLLMRouter is on, deploy/mavend.json sets llm_router true. Rewritten, with the measured numbers for both engines and the note that the classifier is the failure floor and must not be deleted. - CLAUDE.md persona line pointed at a memory file that does not exist, so the rule was nowhere in the repo. Written out, with the three eval checks that enforce it. - MODEL-BAKEOFF: eval-models target does exist now; the routing numbers are the production path, not a bench artifact; the truncated gguf is deleted. Struck through rather than deleted. Docs only, no code touched.

kami added 1 commit 2026-07-31 20:37:31 +02:00

Stop the docs claiming the LLM router is off 0ca5748699

CLAUDE.md's routing section said "llmrouter is wired nil" and called the
classifier cascade the committed default. That stopped being true when the
integration merge landed: voice.go:214 wires pickLLMRouter, DefaultLLMRouter is
on, and deploy/mavend.json sets llm_router true. It is the first thing anyone
reads before touching the router, so it was pointing the next reader at a
wiring job that is already done.

Rewritten to say the LLM router is the default, the classifier is the failure
floor and must not be deleted, and what the two actually measure — 36.8% at
p50 31ms against 67.5%/72.7% at p50 ~2.7s, a trade accepted on purpose. Names
the one thing still open on that path: Confidence is hardcoded 1.0 in
llmrouter.go, so the LLM never asks for clarification (#359).

Also in CLAUDE.md: the persona line pointed at a memory file that does not
exist, so the actual rule was nowhere in the repo. Written out instead —
feminine self-reference, informal singular address, pet names forbidden but his
name allowed — plus the three eval checks that enforce it.

MODEL-BAKEOFF: three claims had gone stale within hours of being written. There
IS a make eval-models target now; the routing numbers ARE the production path,
not a bench artifact waiting on a wiring change; and the truncated 293 MB gguf
is deleted. Struck through rather than removed, since the caveats are part of
how the evening read at the time.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CGeSZxh1DCtRxmFVSYVGvJ

kami merged commit 73d13f1ea6 into master

2026-07-31 20:46:58 +02:00

Sign in to join this conversation.

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: kami/Maven#49