L26 QUERY LOG
Label for the answers and any scaffold changes: AI allowed — keep a record
Label for the diagnosis column: No AI (the failure analysis is your own work)

Settings used: CHUNK_WORDS = ____   TOP_K = ____
(Start a new log section whenever you change a setting.)

For every query, record what was RETRIEVED before you read what was ANSWERED.

Query number and question:
  Document I expect to answer it (written before running):
  Retrieved (rank, score, document, chunk):
    1.
    2.
    3.
  Did the text that answers the question reach the prompt? (yes / no / no answer exists)
  Assistant's answer (paste it, citations included):
  Does the retrieved text support every claim in the answer? (yes / partly / no)
  Failure? If so, which stage: chunk / represent / store / retrieve / assemble / generate
  Evidence for that stage (one or two sentences):

(copy the block above for each query)

ONE CHANGE AND ITS SIDE EFFECT
  Setting I changed (from -> to):
  What I predicted would improve:
  What I predicted might get worse:
  What actually changed (query numbers):

AI-USE RECORD
Label:            AI allowed — keep a record
Assistant:        (name and version, or "local model")
Date(s):
Prompts:          (paste them, or a link to the conversation)
What I kept:
What I changed:   (and why)
What I verified:  (which claims or results I checked, and against what)
