[joyful-ii-V-I]

The published gate count was hand-written at eight sites across three files. Two lanes that
each add one gate both write N+1; git auto-merges the identical text clean, and the tree
then publishes N+1 against a loop of N+2 — with every existing check green, because "my count
equals my own loop" holds on each branch, "my loop equals main's loop" holds after the merge,
and the member sets differ at the same number. It collided seven times in one night on
2026-09-10 and serialised every gate-adding lane.

This makes the number a build product, the way docs/COMMANDS.md and docs/LIMITS.md
already are.

docs/gatecount_build.py derives the count from the single for _g in …; do loop in
test/regression.sh and rewrites all eight sites. --check exits 1 on drift, --root
retargets a synthetic tree, --sites prints the site list.

The marker. Each site carries a comment the generator owns — <!-- gatecount --> in
markdown (invisible when rendered on GitHub) and // gatecount in the deck's JavaScript. A
count claim on an unmarked line is a hand-written count and the generator refuses the tree
rather than leave it behind; a marker that has lost its claim is refused too.

test/gatecountcheck.sh, written before the generator and watched red:

arm	what it proves
A	the generator's count matches an independent derivation of the loop
B	every marked site states what the loop names right now
C	can-go-red on a synthetic tree via --root (never a probe file in this one — that perturbs the crawl other gates measure); a second added gate reds it again
D	a stripped marker is refused in both --check and write mode
E	an unparseable loop is a refusal, not a fabricated number
F	the three hand-kept copies of the site list (generator, gate, manifestcheck) agree

manifestcheck's arms are all kept, none weakened. The generator is how the sites are
written; manifestcheck is what notices one written some other way. Arm (F) closes the gap
those two site lists left — nothing compared them before.

Also proven by hand: editing the deck's stat(s, "578", …) spelling by hand reds arm (B) and
naming the site; stripping a marker from README.md reds (A) and (B) with both claim lines
listed; regenerating clears each.

The merge recipe for a gate-adding lane is now: union the for _g in … sets, run
python3 docs/gatecount_build.py, done. 577 → 578 in this PR was produced by that command,
not by hand.

Registered in the same commit: the loop, binoverridecheck's pinned exemption list (the gate
binds no ripwire binary, so (2b)'s static tell needs no row), docs/README.md's index, the
CONTRIBUTING.md submission checklist, and docs/EVALS.md §6/§8.

Touches no src/ — the binary is unchanged, so --for / --pack-task / the flagless map are
byte-identical by construction.

Local: gatecountcheck, manifestcheck, limitstablecheck, readmedriftcheck,
docscommandscheck, deckcheck, deckclaimcheck, ripwirepubliccheck, gateexitcheck,
gateabilitycheck, pargatescheck, loopconservationcheck, qschemetripcheck,
runhintcheck, testmacrocheck, affectedcheck — all green. binoverridecheck (runs every
non-exempt gate against a broken stub) is left to CI.

The motivating case, observed on a real rebase an hour after this PR opened

A lane rebased onto dc48e1e9. test/regression.sh was the ONLY conflict, and all eight published count sites auto-merged clean to 577 while the resolved loop named 578, with not one conflict marker among them, because main's string and the lane's string were both literally "577 gate scripts". The number was wrong in eight places and the only thing that could see it was manifestcheck.

The sites do not conflict because both sides agree on a stale value, so "resolve the conflicts and you are done" is exactly the reasoning that ships the wrong number. After this PR the recipe is: union the for _g in sets, run python3 docs/gatecount_build.py, done. A marker whose claim was deleted is refused (a claim deleted is a claim drifted).

The one red run, for the record

Shard 2 went red on every platform: forrankordercheck arm (3) pins 4% byte bands on ten --for reference queries over the live repo, and the new generator's function named rewrite became a rank-1 name-exact hit for "compact legend rewrite" (+15%). Renamed to rewrite_marked_sites. That arm measures the corpus, not the tool, and two other queries already sit at +3.5% of the band from ordinary growth; filed as a follow-up.

🤖 Generated with Claude Code

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Advanced

Run ID: 18bedc8d-e112-4f00-add7-8fca35962b00

📥 Commits

Reviewing files that changed from the base of the PR and between ccb5c7e and d46d864.

📒 Files selected for processing (5)

• README.md
• docs/EVALS.md
• docs/gatecount_build.py
• present/deck5_ripwire_build.js
• test/regression.sh

🚧 Files skipped from review as they are similar to previous changes (3)

• present/deck5_ripwire_build.js
• README.md
• docs/EVALS.md

Included review availability: Your plan provides up to 4 included reviews per hour; 1 remains after this review.

---

📝 Summary

Summary by CodeRabbit

• Documentation

  • Updated the published test-gate count from 577/578 to 579 across the README, evaluation documentation, and presentation deck.
  • Added contributor guidance for regenerating gate counts and documenting required markers and workflow.

• New Features

  • Added automatic tooling to derive and update published gate counts from the regression suite.
  • Added validation for stale, missing, or malformed gate-count entries.

• Tests

  • Added coverage for gate-count generation, synchronization, marker enforcement, and failure handling.

Walkthrough

The change derives the published gate count from the regression loop, updates marked documentation and deck sites to 579, adds generator checks, and documents the regeneration workflow.

Changes

Gate count generation and validation

Layer / File(s)	Summary
Derive and validate the gate count docs/gatecount_build.py, test/regression.sh, test/gatecountcheck.sh	The generator derives the count from the absorb loop, validates marked sites, and supports checking or rewriting. The new gate tests drift, malformed loops, missing markers, regeneration, and site-list agreement.
Update published count sites README.md, docs/EVALS.md, present/deck5_ripwire_build.js	Published gate counts change to 579. Markdown and JavaScript sites gain generator-owned markers.
Integrate validation and document regeneration test/binoverridecheck.sh, test/manifestcheck.sh, CONTRIBUTING.md, docs/README.md	Existing checks record the new gate and site-list rules. Contributor documentation describes the generator command and marker requirements.

Estimated code review effort: 3 (Moderate) | ~20 minutes

Sequence Diagram(s)

sequenceDiagram
  participant RegressionLoop
  participant GatecountBuild
  participant PublishedSites
  participant GatecountCheck
  RegressionLoop->>GatecountBuild: derive gate count
  GatecountBuild->>PublishedSites: check or rewrite marked counts
  GatecountCheck->>GatecountBuild: verify drift and refusal cases
  GatecountBuild-->>GatecountCheck: return validation status

Loading

Suggested reviewers: quaterniondrift

Merge Risk: ⚪ Minimal · up to d46d8

The change generates gate counts from the regression loop and validates published sites; current validation reports pass with no actionable merge-blocking risk identified.

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 40.00% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 5 functions across 6 files. (2 skipped: 2…	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly identifies the main change: the published gate count becomes a generated build product and addresses the merge issue that motivated it.
Description check	✅ Passed	The description directly explains the generator, validation gate, marker design, merge failure mode, registered files, and validation results.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

Full details: Docstring Coverage

Explanation

Docstring coverage is 40.00% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 5 functions across 6 files. (2 skipped: 2 unsupported.)

• Fix all pre-merge checks with AI

✨ Finishing Touches 💡 1

📝 Generate docstrings 💡

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch lane/gatecount-generated-2026-09-10

---

_{Comment @coderabbitai help to get the list of available commands.}

[joyful-ii-V-I]

Landed as main 1d11c8d. Loop membership = the 578 members main had at 6f91fed (Dart) plus exactly gatecountcheck = 579, and the eight published sites read 579 because the generator wrote them. For every lane behind this: union your for _g in set with main's, run python3 docs/gatecount_build.py, commit the result. Nobody needs the number.