[hieuphung97]

0.17.3 required a disposition to wait until the reviewer had located the
counterexample's red run in the implementer's dispatch record, with an escape
for a record Orca cannot recover. 0.17.5 added a second escape. A third case was
found and queued.

A Spike measured the lookup instead of extending it again.

What was measured

worker-read across all ten settled dispatches of those deliveries returns one of
two shapes:

• Cursor: source: terminal, 120 lines, truncated: true, cursor: null —
  no way to page backwards.
• Claude Code: source: transcript with a cursor, warning that older messages
  were omitted from the bounded archive.

Searching a Cursor implementer's retained window for its own evidence found the
fixture command 0 times and the failing output 0 times. Those 120 lines are
the worker's closing summary, not the middle of the session where the
counterexample ran.

AGENTS.md pins Cursor for implement. So in this repository the implementer's
dispatch record structurally cannot hold the red run.

Across the five reviews written under the rule, none located it. All five
reproduced the counterexample themselves and said so.

What changes

The lookup is deleted with both escapes. The surviving obligation joins the rule
it always belonged to:

Reproduce, do not accept. Run the gates yourself. A claim you did not
reproduce is not evidence. The reviewer observes the counterexample discriminate
for itself — an implementation that is present, runs, returns a pass, and is
wrong. An instrument red only because the behaviour was absent is not that
observation.

## Review's opening no longer describes worker-read; ## Evidence and the
### Handoff Verification clause already carry that machinery, including what to
do when Orca cannot recover a dispatch item.

This removes the dependency on plane retention, so the queued partial-recovery
case needs no branch.

Size

## Review: 353 → 206 words. It remains 26 words larger than the 180 it
was before the lookup existed — a reduction against the peak, not a return to the
starting point. An earlier draft of this description claimed otherwise; review
caught it, and the decision record states the residue plainly.

Does the shorter text still catch the failure it was written for?

That was the acid test put to review, with instruction not to soften a Blocking
finding to preserve a reduction. Its answer: yes, and more reliably — the reviewer
must observe the discrimination from its own seat rather than search a window that
cannot hold it. It discharged the new obligation in that review, where five prior
reviews could not discharge the old one.

Review found two further errors, both mine

The word-count claim above, and a Context sentence in the decision record that
compressed three releases into "two further deliveries each added a branch". The
correct history is one escape shipped with the rule, one added later, one queued
and never shipped — verified by counting escapes per commit and by
git log --all -S.

Version

0.17.6 in all three coupled sites.

🤖 Generated with Claude Code

https://claude.ai/code/session_01RFPwA3GEzovwEoeohLLPjM