Skip to content

Statement audit 2026-11: restate unclear problems (#301) - #335

Merged
Kyu-seok merged 197 commits into
masterfrom
statement-audit-2026-11
Nov 14, 2026
Merged

Kyu-seok merged 197 commits into
masterfrom
statement-audit-2026-11

Conversation

@Kyu-seok

@Kyu-seok Kyu-seok commented Nov 14, 2026 •

Copy link
Copy Markdown
Owner

Implements spec #301 (statement audit 2026-11): restates the unclear problems ticket by ticket, one Restatement commit per problem, each ticket merged with --no-ff. #302 (the tracer) shipped separately as #334.

Closes #301
Closes #303
Closes #304
Closes #305
Closes #306
Closes #307
Closes #308
Closes #309
Closes #310
Closes #311
Closes #312
Closes #313
Closes #314
Closes #315
Closes #316
Closes #317
Closes #318
Closes #319
Closes #320
Closes #321
Closes #322
Closes #323
Closes #324
Closes #325
Closes #326
Closes #327
Closes #328

What users will notice

  • Renamed Slugs: old links redirect through Former Slugs. Editor drafts saved under an old Slug stay behind, and new Solution Commits go to the new folder.
  • Renamed methods: earlier code calling the old names no longer compiles if resubmitted. Stored Verdicts are untouched.
  • Production comparison not done (scripts/prod-text.sh is denied to agents): any statement edited through the admin page in production is overwritten by these migrations. Please run scripts/prod-text.sh <old-slug> for the problems below before merging.

Per ticket

  • Rename misleading titles (1/2) #303: V320–V324. longest-quiz-streak→longest-rising-quiz-scores (longestRisingScores); best-boba-streak→max-product-slice (maxProduct(nums)); top-raid-score→kth-highest-raid-score (kthHighestScore); unique-genre-window→longest-window-with-at-most-k-genres (longestKGenreWindow; Walkthrough moved+regenerated; removed tracked .pyc — other .pyc in scripts/walkthroughs/pycache tracked too); longest-unique-treats→longest-two-flavor-stretch (longestTwoFlavorStretch; constraint loosened to 0<=treats[i]<=10^5, audit missed). All ACCEPTED 4 langs + Run. Redirect via API only. Prod comparison not done.
  • Rename misleading titles (2/2) #304: V330–V334. best-skin-flip-day→skin-resale-with-cooldown (maxProfit(prices)); validate-homework-id→match-dot-star-pattern (isMatch); grocery-bag-triple→combine-bags-to-match-target (canMatchTarget(bags,target); loosened 0<=bags[i][j] for hidden case 12); verify-pizza-rotation→non-decreasing-with-one-change (canBeNonDecreasing(nums)); calm-phone-number→calm-number (isCalm(reading)). All 4 langs ACCEPTED judge-local + Run. Redirect checked via API only. Prod comparison not done.
  • Restate hard-to-read statements (severity 2) (1/10) #305: V340–V344, no slug changes. snapshot-town-map (cloneMap; empty map Example 3); find-locker-number (findLocker(lockers,target)); merge-leaderboard-runs (mergeRuns); detect-portal-loop (hasPortalLoop; ListNode.val); verify-task-order (canFinishAllTasks(numTasks,prereqs)). All ACCEPTED 4 langs + Run. Prod comparison not done.
  • Restate hard-to-read statements (severity 2) (2/10) #306: V350–V354, no slug changes. word-break-chat (canSplitMessage(message, words); ko title 채팅 메시지 단어 나누기); finish-the-level (canFinishLevel(jumps)); lonely-candy (findLonelyCandy(candies); Walkthrough regenerated); next-speaker-order (nextLineup(lineup); Example 1 lists the 6 orders rather than steps, to avoid giving away the algorithm; ids may repeat); eval-postfix-tape (evalPostfix(tokens); added 32-bit intermediate constraint, true for all cases). All ACCEPTED 4 langs + Run. Prod comparison not done.
  • Restate hard-to-read statements (severity 2) (3/10) #307: V360–V364. fix(problem): clone-quest-log Java Driver never defined Node (every Java submission COMPILATION_ERROR) — fixed. interleave-clip-chain (interleaveClips; ko 클립 체인 교차하기); clone-quest-log (cloneQuestLog; Node.random→Node.related); rebuild-boss-tree→rebuild-tree-from-traversals (rebuildTree; slug renamed though ticket didn't ask — "Boss" theme word); wildcard-lexicon (python add_word→enroll; call-by-call table); dual-waterbowl-flow (text only; corrected wrong Example 1 explanation). All ACCEPTED (dual-waterbowl-flow Python only, text-only) + Run. Prod comparison not done.
  • Restate hard-to-read statements (severity 2) (4/10) #308: V370–V374, no slug changes. order-group-tasks (findTaskOrder(numTasks,…); title kept, "Group" explained; ko 조별 과제 순서 정하기); rotting-donuts (minutesUntilStale(tray); return 0 when nothing fresh; Walkthrough regenerated); count-cipher-decodings (countDecodings(digits); ko 암호 해독 방법 세기); count-doodle-palindromes (countPalindromicSubstrings(doodle); ko 낙서 회문 세기); desk-notes-overlap (longestCommonSubsequence(left,right)). All ACCEPTED 4 langs + Run. Prod comparison not done.
  • Restate hard-to-read statements (severity 2) (5/10) #309: V380–V384, no slug changes. online-friend-mask (countOnlineFriends(friendMask); constraint fixed to 0<=friendMask<2^32, Java/C++ negative int note; ko 접속 중인 친구 수); repeat-attendance-codes (text only; empty-list Example 3; Walkthrough regenerated); four-tray-sum (fourTraySum; constraints 0<=len<=200, ±10^9; 64-bit warning); pair-lunchbox-snacks (minLunchboxes(weights)); group-car-packs (countCarPacks; catching up at booth counts). All ACCEPTED 4 langs + Run. Prod comparison not done.
  • Restate hard-to-read statements (severity 2) (6/10) #310: V390–V394. merged-price-median (mergedPriceMedian(pricesA,pricesB); lists are price changes, negatives); min-daily-brew→min-daily-brew-capacity (minDailyCapacity); running-median-timer (text only; call tables; num bound fixed to ±10^5; title kept since running-median is own Former Slug); space-cookie-bakes (spaceBakes); snack-combo-picks (listSnackCombos; includes empty combo, any order). All ACCEPTED 4 langs + Run. Prod comparison not done.
  • Restate hard-to-read statements (severity 2) (7/10) #311: V400–V404, no slug changes. cafe-price-ratios (text only; "items"; rounded to 5 decimals; consistent-ratios constraint); cheapest-courier-route (cheapestRoute(…, maxStops); dropped false src!=dst; Walkthrough regenerated); rising-water-path (earliestCrossingTime(elevation); loosened bounds: <=3600, heights may repeat); recover-tier-order (recoverTierOrder(tags); "" for both contradictions; size constraints added); min-stamina-checkpoints (minStaminaToFinish(cost); bound <=1000). All ACCEPTED 4 langs + Run. Prod comparison not done.
  • Restate hard-to-read statements (severity 2) (8/10) #312: V410–V414. split-ticket-pile→split-tickets-into-equal-sums (canSplitEqually); count-square-candy-bags→fewest-square-candy-bags (fewestBags(candies)); split-candy-bar→split-candy-bar-for-max-product (maxSplitProduct(length); bound 60→58, int overflow at 59, audit missed); ticket-plus-or-minus (countSignAssignments; 0 counts twice); best-balloon-pop-order→max-balloon-pop-points (maxPopPoints). All ACCEPTED 4 langs + Run; redirects via API. Prod comparison not done.
  • Restate hard-to-read statements (severity 2) (9/10) #313: V420–V424. count-coupon-ways (text only; amount=0 returns 1); longest-level-up-path (longestIncreasingPath; length counts tiles); balance-print-queue→balance-brackets-with-wildcards (canBalance(brackets)); battery-drain-power→fast-exponentiation (power(base, exponent); loosened exponent to full int range and allowed 0^0=1, audit missed both); power-of-two-tank (isPowerOfTwo(n); bound fixed to full int range, audit missed). All ACCEPTED 4 langs + Run; redirects via API. Prod comparison not done.
  • Restate hard-to-read statements (severity 2) (10/10) #314: V430–V432, no slug changes. compare-loadout-bits (params x,y→leftFlags,rightFlags; bound fixed to 2^31-1, audit missed); chat-ring-buffer (text only; fixed-capacity queue, no overwrite; call tables); snack-stock-map (text; 0<=key,value<=10^6; banned built-ins named). fix(problem) 2520919: snack-stock-map Java Driver called List.lookup since 420a642 (Apr 2026) — every Java submission was a compile error. All ACCEPTED 4 langs + Run. Prod comparison not done.
  • Restate statements needing polish (severity 1) (1/5) #315: V450–V455, no slug changes. reverse-linked-list (text; ListNode.val); cafe-stop-orders (listVisitOrders(stops); loosened to 0<=length, empty→[[]]); voucher-price-combos (text; 1<=prices[i]; any order; []); top-espresso-machines (text; unique-answer guarantee); product-of-all-other-elements (text; no-division rule up front); longest-seat-run (longestSeatRun(seatIds); duplicates count once, empty→0). All ACCEPTED 4 langs + Run. Prod comparison not done.
  • Restate statements needing polish (severity 1) (2/5) #316: V460–V465, no slug changes. rotate-pixel-art (text); trapping-rain-water (text; 0<=n empty→0; height bound 10^8→10^4 to avoid int overflow, audit missed); list-bracket-pairings (listBracketPairings(pairCount); Korean definition rewritten); slowest-snack-rate (minEatingRate(piles, hours); hours<=10^9, audit missed); best-skill-tree-path (maxPathSum; ko title 경로); fewest-tile-jumps (minJumps(maxJump); fixed Example 2 "tile 7", audit missed). All ACCEPTED 4 langs + Run. Prod comparison not done.
  • Restate statements needing polish (severity 1) (3/5) #317: V470–V475, no slug changes. fix(problem): timestamp-value-store Java Driver called List.readAt since 420a642 — every Java submission failed to compile. peak-table-demand (minTables(visits); 0<=length, empty→0); shortest-cookie-span (shortestCookieSpan(trays); ko "target 이상"); timestamp-value-store (text); find-menu-calorie (findCalorie(menu)); lru-cache (text; put updates and counts as use — no case tests it); binary-tree-codec (text; level-order null notation). All ACCEPTED 4 langs + Run. Prod comparison not done.
  • Restate statements needing polish (severity 1) (4/5) #318: V480–V485, no slug changes. populating-next-pointers (text; node count [0,4095], empty→null); sum-root-to-leaf (text; 32-bit guarantee as constraint); micro-feed (text; recency = call order; call tables); two-shift-baristas (text; shift may be empty); guild-chat-relay (minutesToReachAll; 0<=links.length, 1<=src<=n); redundant-friendship (text; last cycle edge rule). All ACCEPTED 4 langs + Run (V484 wording fixup after Run check, tests ran on final). Prod comparison not done.
  • Restate statements needing polish (severity 1) (5/5) #319: V490–V494, no slug changes. dog-street-walk (planWalk(routes); zone-code constraint fixed to 1–3 letters, hidden case 11); check-interleaved-inputs (isInterleaving; all-empty rule in prose; ko title 두 입력 교차 병합 검사); can-attend-bookings (text; any order); zero-out-rows-cols (zeroOutRowsAndCols(board); ko title 0이 있는 행과 열 지우기); multiply-as-text (ko only; 세로 곱셈 typo). All ACCEPTED 4 langs + Run. Prod comparison not done.
  • Rename leftover Starter Code names (1/9) #320: V500–V504 (two-menu-sum no migration). valid-brackets (isValidBrackets(brackets); full rule); mirror-menu-tree (mirrorTree; empty tree stated); zero-sum-trio (zeroSumTrios; any order, values may repeat); valid-slogan-palindrome (isPalindrome; ASCII 32–126); two-menu-sum (findMenuPair; text unchanged); biggest-locker-poster (largestPosterArea(lockers); height bound 8*10^4→10^4, int overflow, audit missed). All ACCEPTED 4 langs + Run. Prod comparison not done.
  • Rename leftover Starter Code names (2/9) #321: V510–V514 (same-skill-trees no migration). find-menu-price (hasPrice(menu)); find-rotated-low-score (findLowestScore(scores); rotation defined, Example 3 contradiction fixed); deepest-skill-chain (maxSkillDepth); same-skill-trees (isSameTree, names only); valid-score-bst (isValidBST; dropped false "distinct"; bound widened to 32-bit for hidden [2147483647], audit missed); count-desk-clusters (countClusters(board)). All ACCEPTED 4 langs + Run. Prod comparison not done.
  • Rename leftover Starter Code names (3/9) #322: V520–V525, no slug changes. count-level-hops (countHopSequences; n<=40→45 for hidden n=45); pick-snack-haul (maxSnackHaul(snacks)); fewest-candy-bags (fewestBags(bagSizes, quota); quota 0→0 bags); peak-score-streak (maxScoreStreak; |nums[i]| 10^8→10^4, int overflow, audit missed); missing-sock-number (smallestMissingNumber(socks)); move-voided-scans (moveVoidsToEnd(scans)). All ACCEPTED 4 langs + Run. Prod comparison not done.
  • Rename leftover Starter Code names (4/9) #323: V530–V533, no slug changes. find-name-anagram (containsNameAnagram, no migration); peak-combo-window (peakComboPerWindow(combo)); days-until-warmer (daysUntilWarmer(readings), no migration); nth-lowest-score (nthLowestScore); right-side-skill-view (rightSideView; Example 3 dropped as near-duplicate, sample 3 still a Run sample); longest-folder-path (longestPathEdges). All ACCEPTED 4 langs + Run. Prod comparison not done.
  • Rename leftover Starter Code names (5/9) #324: V540–V545, no slug changes. flip-phone-codes (letterCombinations; only Python differed); count-voice-parties (countParties(relays)); pick-circular-donuts (maxTastiness(tastiness); single donut can be taken); count-dungeon-paths (countPaths(rows, cols); bound 80→15, int overflow, audit missed); slot-a-meeting (insertMeeting(meetings, newMeeting); touching defined as merging — no case tests it); spiral-shelf-walk (spiralOrder(shelf)). All ACCEPTED 4 langs + Run. Prod comparison not done.
  • Rename leftover Starter Code names (6/9) #325: V550–V555. majority-boss-vote→majority-vote (majorityVote(votes); Walkthrough moved+regenerated; old pyc left tracked); sort-latte-squared (sortedSquaredDrifts(drifts)); flip-chunks-of-k (flipChunks; "chunk"); count-top-heroes (countDominantHeroes); balanced-hamster-tree (isBalanced; height defined); zigzag-yearbook-order (zigzagRows; empty tree Example 3). All ACCEPTED 4 langs + Run; redirect via API. Prod comparison not done.
  • Rename leftover Starter Code names (7/9) #326: V560–V565, no slug changes. list-skill-build-paths (listBuildPaths; any order, empty list); flatten-agenda-tree (flattenAgenda; duplicate Example 3 dropped; ko title 안건 트리 펼치기); task-cooldown-planner (minPrintSlots; n=0 Example 3; ko title 작업 쿨다운 플래너); menu-combos-tab (listTabCombos(pastries, tab)); restore-wifi-ip (restoreIpAddresses(digits); "any order" since Drivers sort; ko title Wi-Fi IP 복원하기); capture-enclosed-tiles (captureEnclosedTiles; returns nothing). All ACCEPTED 4 langs + Run. Prod comparison not done.
  • Rename leftover Starter Code names (8/9) #327: V570–V573, no slug changes. shortest-dungeon-path (shortestPathLength(board), no migration); connect-water-bowls (minHoseLength(bowls); coords 10^8→10^5, int overflow, audit missed; ko title 물그릇 연결); nth-unlockable-level (nthUnlockableLevel); count-hint-matches (countHintMatches(notes, target); longer target→0, no case tests it); group-study-cards (canGroupCards(cards), no migration); rebuild-tournament-line (rebuildLine(players)). All ACCEPTED 4 langs + Run. Prod comparison not done.
  • Rename leftover Starter Code names (9/9) #328: V580–V581, no slug changes. fix(problem): versioned-cells Java Driver called List.readAt since 420a642 — every Java submission failed to compile. cover-aquarium-zones (minPellets(zones), no migration); palindrome-episode-number (isPalindrome(episode)); reverse-savefile-flags (reverseFlags; unsigned 32-bit result; Example 3 past signed range); versioned-cells (python snapId; Driver locals). All ACCEPTED 4 langs + Run. Prod comparison not done.

Beyond the tickets

  • Four Java Drivers fixed (fix(problem) commits): clone-quest-log never defined Node; snack-stock-map, timestamp-value-store and versioned-cells called renamed methods on Java Lists since 420a642 (April). Every Java submission to these four failed to compile in production until this merges.
  • Bounds the audit missed: stated maximums that broke a hidden case or overflowed the int return type were corrected (loosened to fit a fair case, or tightened where every case stays far inside). No Test Case changed.
  • Six Slug renames the tickets didn't tag (kept on review): rebuild-boss-tree, majority-boss-vote, split-candy-bar, balance-print-queue, split-ticket-pile, min-daily-brew. Old links redirect.
  • Two Examples dropped (right-side-skill-view, flatten-agenda-tree): each duplicated another, and its sample case can't change in a Restatement; the sample still runs.
  • Review fixes: five Python Starter Codes moved from snake_case to the camelCase the other languages use (count-cipher-decodings, count-doodle-palindromes, desk-notes-overlap, order-group-tasks, rotting-donuts); scripts/walkthroughs/__pycache__ untracked and ignored; the restate-problem skill gained the audit's lessons (bounds vs return type, judge all 4 languages before renaming, C++14 references).
  • Java, JavaScript and C++ reference solutions for every restated problem, so judge-local.sh covers all 4 languages.

Checks

Each ticket: judge-local.sh ACCEPTED in all 4 languages on every case, the Run endpoint on every sample, the Former Slug redirect through the API, com.codebite.problem.* and the Driver/Starter Code tests. Each merge: the full backend, worker and frontend checks via scripts/merge-ticket.sh.

🤖 Generated with Claude Code

Kyu-seok and others added 16 commits November 15, 2026 00:45
New Slug longest-rising-quiz-scores, retitled "Longest Rising Quiz
Scores". "Streak" read as consecutive quizzes; the task is a subsequence that may
skip quizzes. Starter method longestBrightRun -> longestRisingScores, in
all 4 Starter Codes and Drivers. Sample 3 becomes Example 3. Java,
JavaScript and C++ reference solutions added.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
"Streak" hid that the task is the largest product, and the foot-traffic
story made multiplying counts meaningless. Starter method bestFlowStreak
(max_product in Python) -> maxProduct, parameter traffic -> nums, in all
4 Starter Codes and Drivers. Fixes the Korean Example 2 that said the
opposite of what it meant. Java, JavaScript and C++ reference solutions
added.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
"Top" read as the maximum; the task is the k-th highest. Starter method
findBrightestTargets -> kthHighestScore, in all 4 Starter Codes and
Drivers. Java, JavaScript and C++ reference solutions added.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
State the task first (a deep copy of the map containing node), say the
adjacency list is only how the examples write a map, and show the empty
map as Example 3. Starter method snapshotMeshLayout -> cloneMap in all 4
Starter Codes and Drivers; reference solutions in all 4 languages.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
One theme and one name per thing: the array is lockers (was shelf) and
the number target (the text said query) in text, constraints and Starter
Code. Starter method locateCallNumber -> findLocker in all 4 Starter
Codes and Drivers; reference solutions in all 4 languages.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
State the merge first, explain the examples without the friends they
never introduced, and say run times (기록) in Korean too, not lap times.
Starter method combineOvenRotations -> mergeRuns in all 4 Starter Codes
and Drivers; reference solutions in all 4 languages.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
New Slug longest-window-with-at-most-k-genres, retitled "Longest Window
With At Most K Genres": "Unique" suggested all-distinct tags, but the
task allows up to k distinct genres. Starter method countUniqueTopics
(k_distinct_window in Python) -> longestKGenreWindow, in all 4 Starter
Codes and Drivers. Sample 3 becomes Example 3. The Walkthrough and its
tracer move to the new Slug and are regenerated. Java, JavaScript and
C++ reference solutions added.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
New Slug longest-two-flavor-stretch, retitled "Longest Two-Flavor
Stretch": "Unique" suggested all-distinct treats, but the task allows at
most two flavors. Starter method longestCleanRun (fruit_picker in
Python) -> longestTwoFlavorStretch, in all 4 Starter Codes and Drivers.
The constraint 0 <= treats[i] < treats.length, false for several cases,
is loosened to 0 <= treats[i] <= 10^5. Java, JavaScript and C++
reference solutions added.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The text named detectTransferLoop(head, pos) while the function takes
only head; pos is now explained only as how an example builds the loop.
The constraint names ListNode.val, the Starter Code's class. Starter
method detectTransferLoop -> hasPortalLoop in all 4 Starter Codes and
Drivers; reference solutions in all 4 languages.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
State the task first and name the task count numTasks (was numScenes,
a film-shoot leftover) in text, constraints and Starter Code. Starter
method verifyShootPlan -> canFinishAllTasks in all 4 Starter Codes and
Drivers, whose locals now read numTasks and prereqs; reference solutions
in all 4 languages.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
New Slug: skin-resale-with-cooldown.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
New Slug: match-dot-star-pattern.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
New Slug: combine-bags-to-match-target.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
New Slug: non-decreasing-with-one-change.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Kyu-seok and others added 13 commits November 15, 2026 00:52
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…305)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Starter method splitHeadlineTokens -> canSplitMessage, parameters
headline -> message and topics -> words; Example 3 explained step by step;
Korean title matches the English one.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A 0 no longer reads as a pit you can't land on. Starter method
canFinishLine -> canFinishLevel, parameter slack -> jumps.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Starter method findUnpairedFrame -> findLonelyCandy, parameter
frames -> candies; Walkthrough regenerated.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Starter method advanceShotOrder -> nextLineup, parameter shotOrder ->
lineup; lexicographic order defined and Example 1 lists the arrangements;
repeated ids stated. Walkthrough regenerated.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Starter method foldBaleOps -> evalPostfix, parameter baleTape ->
tokens; operand order of - and / stated; 32-bit intermediate bound stated.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The Java Starter Code shows Node only in a comment and the Driver never
defined it, so every Java submission failed to compile. The Driver now
declares Node as the C++ Driver does. Adds the Java reference solution
that found it.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Kyu-seok and others added 23 commits November 15, 2026 02:33
Retitled "Majority Vote": "Boss" was never explained and "Return that id"
named ids the statement never introduced. Starter method majorityHeadline
-> majorityVote in all 4 Starter Codes and Drivers; the Walkthrough moves
with the Slug.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The statement now opens with the task. Starter method orderTankDeviations
(tanks, from an earlier theme) -> sortedSquaredDrifts in all 4 Starter
Codes and Drivers.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Drops the power-ups theme and says "chunk" as the title does. Starter
method flipSorterSegments (reverse_k_group in Python) -> flipChunks in
all 4 Starter Codes and Drivers.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The statement now opens with the task and each example names the path
maximum a hero is compared with. Starter method tallyDominantBlocks
(blocks, from an earlier theme) -> countDominantHeroes in all 4 Starter
Codes and Drivers.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Drops the pet-shop opening, starts with the task and says what a
subtree's height counts. Starter method checkHubEvenness -> isBalanced
in all 4 Starter Codes and Drivers.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Drops the yearbook-photo opening, starts with the task, and makes the
empty-tree sample Example 3. Starter method alternateWardSweep (wards,
from an earlier theme) -> zigzagRows in all 4 Starter Codes and Drivers.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The method was coverAllWindows in Java, JavaScript and C++ and
find_min_arrows in Python, names left over from earlier themes; the
statement drops pellets on zones, so all four Starter Codes and Drivers
now call minPellets(zones). The statement names no method, so its text
and the Test Cases stay. Reference solutions in all four languages.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The title's episode number never appeared in the statement, which read
an integer `code` and called the method isReversibleCode. The statement
now reads it as an episode number, and all four Starter Codes and
Drivers call isPalindrome(episode). Title, Slug and Test Cases stay.
Reference solutions in all four languages.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The description named the method reversePrinterFlags, a printer left
over from an earlier theme, and asked for "a `long`", which only Java
has. It now states the task first, names no method and says the result
is an unsigned 32-bit value; sample case 3 becomes Example 3, a result
past the signed 32-bit range. All four Starter Codes and Drivers call
reverseFlags(statusWord). Title, Slug and Test Cases stay. Reference
solutions in all four languages.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Commit 420a642 renamed the method get to readAt across the problem's
files and also rewrote the Java Driver's ops.get(idx) and
argsList.get(idx), which are List calls: List has no readAt, so every
Java submission has failed to compile since. Both calls are get again.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The statement and three Starter Codes name readAt's second parameter
snapId; the Python Starter Code called it snap_id, and now says snapId
too. The Drivers' instance, snapArr, is cells, as in the statement's
call trace. The statement and Test Cases stay. Reference solutions in
all four languages.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The tracers set sys.dont_write_bytecode, so the tracked .pyc files were
stale leftovers that nothing reads.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Five problems the audit touched named the Python method (and, for
order-group-tasks, a parameter) in snake_case while Java, JavaScript and
C++ use camelCase. Rename the Python Starter Code, the Python Driver's
call and locals, and the reference solution to the other three's names:

- count-cipher-decodings: count_decodings -> countDecodings
- count-doodle-palindromes: count_palindromic_substrings ->
  countPalindromicSubstrings
- desk-notes-overlap: longest_common_subsequence -> longestCommonSubsequence
- order-group-tasks: find_task_order -> findTaskOrder, num_tasks -> numTasks
- rotting-donuts: minutes_until_stale -> minutesUntilStale (also in the
  Walkthrough tracer's docstring; its output is unchanged)

No statement names a Python method, so no migration.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Fold the durable lessons from spec #301's tickets into the skill: check
every bound against every case and the return type, judge all 4
reference solutions on the unchanged problem before renaming, grep all 4
Starter Codes for the method, the reference solution conventions per
Language (C++14 above all), skills in the rename grep, Java migrations,
whole-word perl renames, port and Flyway pitfalls of branch servers.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…6-11

# Conflicts:
#	.claude/skills/restate-problem/SKILL.md
@Kyu-seok
Kyu-seok merged commit 3804db6 into master Nov 14, 2026
4 checks passed
@Kyu-seok
Kyu-seok deleted the statement-audit-2026-11 branch November 14, 2026 22:03
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment