fix(knowledge): matched was an opinion, and eleven tables that do not exist got it

[sshlg]

sync_status is the word a customer reads on the code↔DB screen. matched says your code and your database agree about this table.

Measured in production, 2026-08-27:

db_index rows for  bs  ls  os  zes  active  unlimit  esim
                   clones  interfaces  price  contract      →  0

code_db_sync status for those same eleven                   →  matched, all of them

Eleven tables that are not in the database were reported as agreeing with it.

The mechanism, not a guess

_match_tables builds the code-only tail with _make_matched(nm, "", …) — db_context empty, because there is no DB side. That struct goes to the LLM, which returns a sync_status.

The deterministic SYNC-L5 override fires only when both column sets are non-empty — which a table with no DB side can never satisfy. So effective_status = analysis.sync_status stood, and the model's answer became the customer's fact.

The order that replaces it

resolve_sync_status():

1. Structure. No DB side → code_only. No code side → db_only. A missing side is a fact; no model can overturn it and none should be asked to.
2. SYNC-L5. The column sets settle matched versus mismatch.
3. The model — only where both sides exist with columns unknown on one of them. That is where its reading is the only reading available.

The fix takes judgement away from where it does not belong and leaves it where it does. Verified on the real names:

bs ls os zes active unlimit esim clones interfaces price contract
  LLM said matched  →  code_only

a genuine pair, code-only column present   →  mismatch
a genuine pair, columns agree              →  matched

Tests

23, with the eleven production names verbatim. It never raises: resolve_sync_status runs inside the sync pipeline, and losing a table's row to an exception is worse than keeping a status that is merely uncertain.

Suite: 6945 passed, 4 skipped, 1 xfailed.

This is the third defect in this feature found in one sitting — the other two being invented table names (#234) and a guard that did not apply on load (#235). All three were found by measuring production rather than reading the code.

🤖 Generated with Claude Code

https://claude.ai/code/session_01XtEZboEDbEXLYrLf1owRXB