data/opus-28.jsonl

{"prompt": "could you plan our live-ops setup for the mobile game? we want to run limited-time events, tune economy values, and A/B test the tutorial without shipping a build. currently everything is baked into the client and a change means a two week store review", "purpose": "planning", "secondary": null, "mixed": false, "difficulty": 0.9, "slice": "core", "lang": "en"}
{"prompt": "remote config service for the game — values fetched at launch, cached, with a version so we can tell which config a session ran", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "live-ops console: active events, the config values with their defaults, and a diff preview before publishing", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "the config fetch blocks the loading screen with no timeout", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.3, "slice": "core", "lang": "en"}
{"prompt": "our economy values are scattered across 30 scriptableobjects, prefabs and a few hardcoded constants. get them all into one config source, same balance in the test scenarios", "purpose": "refactor", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "a config change intended for one region applied globally and the economy was broken for four hours", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "would you explain how a config value reaches the gameplay code — i want to know if a mid-session change can take effect and cause inconsistency", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "live-ops runbook: publishing a config change, the staged rollout, how to verify it landed, and the rollback", "purpose": "writing", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "limited-time event scheduling with start and end in the player's timezone, and a grace period for players mid-session at the end", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "the event countdown in the ui goes negative after the event ends", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.3, "slice": "core", "lang": "en"}
{"prompt": "event ui: banner on the home screen, the reward track, progress, and a clear 'ends in' that's honest about timezones", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "players in some timezones get the event 12 hours late and complain about missing rewards", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "please plan the analytics we need to actually tune the economy — what events, at what granularity, and the queries we'd run to answer 'is this event healthy'", "purpose": "planning", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "economy telemetry: every currency grant and spend with the source, batched and sent reliably even if the app is killed", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "our currency spend events are emitted from 40 call sites with inconsistent source labels. one emission point, and map the historical labels", "purpose": "refactor", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "total currency granted exceeds total spent plus balances by about 4%, which shouldn't be possible", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.9, "slice": "core", "lang": "en"}
{"prompt": "review our telemetry for events that could be spoofed by a modified client, and which of our economy decisions depend on them", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "event taxonomy doc for the game: every telemetry event, its properties, when it fires, and who owns it", "purpose": "writing", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "boundary", "lang": "en"}
{"prompt": "economy dashboard: sources and sinks per currency per day, a net flow chart, and the top spend categories", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "the sinks chart is stacked but the colors are recycled so two categories look identical", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.2, "slice": "boundary", "lang": "en"}
{"prompt": "our retention query defines day-1 retention as 'played again within 24h' in one dashboard and 'played on the next calendar day' in another. pick one, document it, and say which numbers change", "purpose": "refactor", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "the retention numbers in our weekly report don't match the ones in the publisher's dashboard", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "explain what our cohort definition does with players who install and never open the game", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "metric definitions doc for the whole team: dau, retention, arpdau, conversion, each with the exact sql behind it", "purpose": "writing", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "boundary", "lang": "en"}
{"prompt": "ok next", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.3, "slice": "vague-eval", "lang": "en"}
{"prompt": "here's the query plan for our slowest dashboard tile, i've been staring at it for an hour:\n\nGroupAggregate  (cost=2841190.12..2914882.44 rows=412 width=48) (actual time=18422.114..18980.221 rows=390 loops=1)\n  Group Key: d.day, p.platform\n  ->  Sort  (cost=2841190.12..2860112.88 rows=7569104 width=32) (actual time=18410.882..18804.119 rows=7568112 loops=1)\n        Sort Key: d.day, p.platform\n        Sort Method: external merge  Disk: 412880kB\n        ->  Hash Join  (cost=88214.00..1980112.44 rows=7569104 width=32) (actual time=412.118..12088.441 rows=7568112 loops=1)\n              Hash Cond: (s.player_id = p.player_id)\n              ->  Seq Scan on sessions s  (cost=0.00..1412880.04 rows=7569104 width=24) (actual time=0.041..8104.882 rows=7568112 loops=1)\n                    Filter: ((started_at >= '2026-07-01'::date) AND (started_at < '2026-08-01'::date))\n                    Rows Removed by Filter: 84120044\n              ->  Hash  (cost=71204.00..71204.00 rows=1360800 width=16) (actual time=411.002..411.003 rows=1360800 loops=1)\n                    ->  Seq Scan on players p  (cost=0.00..71204.00 rows=1360800 width=16)\nPlanning Time: 0.884 ms\nExecution Time: 19012.440 ms", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "pasted-context", "lang": "en"}
{"prompt": "index on sessions(started_at) and a rewrite so the aggregate doesn't spill 400mb to disk", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "work_mem is 4MB on the analytics role", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.3, "slice": "boundary", "lang": "en"}
{"prompt": "daily pre-aggregated sessions table so the dashboard doesn't scan raw sessions, refreshed incrementally", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "dashboard tiles that load independently with a skeleton each, so one slow tile doesn't hold the page", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "review the other dashboard queries for the same full-month scan of raw sessions", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "write the analytics query guidelines: use the aggregates, when a raw scan is acceptable, and the row limits we enforce", "purpose": "writing", "secondary": null, "mixed": false, "difficulty": 0.5, "slice": "boundary", "lang": "en"}
{"prompt": "plan how we stop the analytics workload from affecting production — a replica, a separate warehouse, or query limits", "purpose": "planning", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "a badly written dashboard query took the production database to 100% cpu for 8 minutes", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "our 30 dashboard queries each embed the same date-dimension logic with slightly different week boundaries. one date dimension table, and report which weeks shift", "purpose": "refactor", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "the week starts on sunday in three queries and monday in the rest", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.3, "slice": "boundary", "lang": "en"}
{"prompt": "explain how our incremental aggregate handles late-arriving sessions from offline play", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "aggregate refresh that handles late data by reprocessing a trailing 7 day window", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "yesterday's numbers change slightly every time you reload the dashboard", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "data freshness indicator on each dashboard tile so people know how current the number is", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.5, "slice": "core", "lang": "en"}
{"prompt": "the freshness indicator shows the query time not the data time", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.3, "slice": "boundary", "lang": "en"}
{"prompt": "plan the semantic layer — one place where a metric is defined, consumed by the dashboards and the notebooks alike", "purpose": "planning", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "plan the analytics warehouse move then build the session aggregation", "purpose": "planning", "secondary": "backendImpl", "mixed": true, "difficulty": 0.8, "slice": "mixed", "lang": "en"}
{"prompt": "figure out why yesterday's numbers drift and then write it up for the analysts", "purpose": "debugging", "secondary": "writing", "mixed": true, "difficulty": 0.7, "slice": "mixed", "lang": "en"}
{"prompt": "onward then", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.3, "slice": "vague-eval", "lang": "en"}
{"prompt": "our spa's interaction latency is terrible and i can't get a clear picture. clicking a row in the main table takes 800ms to show anything, typing in the filter drops frames, and the profiler is a wall of react internals. i want the plan for actually diagnosing and fixing this, with a way to measure it in the field not just on my laptop", "purpose": "planning", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "field performance measurement — INP and long task attribution reported per interaction type, so we know which interactions are slow for real users", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "the row click handler does a synchronous json parse of a 4mb payload", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.4, "slice": "boundary", "lang": "en"}
{"prompt": "our table rerenders every row on every keystroke in the filter because the row component takes a new callback each render. memoize properly, same rendered output", "purpose": "refactor", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "typing in the filter drops to 8fps once the table has 500 rows", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "review our data fetching for waterfalls — i think the detail panel waits for three sequential requests", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "frontend performance guidelines: our interaction budget, the patterns that break it, and how to profile before you optimize", "purpose": "writing", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "row detail panel that shows immediately from data we already have and fills in the rest", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "the detail panel refetches data it already received in the list response", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.4, "slice": "boundary", "lang": "en"}
{"prompt": "performance regression checks in CI using a scripted interaction trace, failing on a budget breach", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "the perf check runs on an unthrottled cpu so it never catches anything", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.3, "slice": "boundary", "lang": "en"}
{"prompt": "plan the bundle diet — we ship 2.8mb of javascript and i suspect half of it is three date libraries and an icon set", "purpose": "planning", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "our app imports three date libraries because different teams added their own. standardize on one, identical formatted output everywhere", "purpose": "refactor", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "the icon library is imported wholesale so all 1400 icons ship", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.3, "slice": "core", "lang": "en"}
{"prompt": "code splitting by route with prefetch on hover for the likely next routes", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "the bundle analyzer shows a 400kb chunk with no obvious source", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "explain which of our dependencies pull in polyfills for browsers we don't support", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "bundle budget documentation: the per-route limits, how they're enforced, and what to do when you need to exceed one", "purpose": "writing", "secondary": null, "mixed": false, "difficulty": 0.5, "slice": "boundary", "lang": "en"}
{"prompt": "bundle size tracking per PR with a comment showing the delta per chunk", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "the size report compares against the wrong base branch so every PR looks like it adds 2mb", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.5, "slice": "core", "lang": "en"}
{"prompt": "pouvez-vous auditer notre bundle et me dire ce qui pourrait être supprimé sans rien casser ? juste une analyse", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "fr"}
{"prompt": "plan how we handle the 8000-row table people keep pasting into the filter — virtualization, server-side filtering, or telling them not to", "purpose": "planning", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "server-side filtering and sorting for the main table, with the client keeping the current ux including the instant-feeling filter", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "the filter debounce is 0 so every keystroke is a request", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.2, "slice": "core", "lang": "en"}
{"prompt": "sure, carry on", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.3, "slice": "vague-eval", "lang": "en"}
{"prompt": "would you propose the design for player support tooling? our support team can currently see nothing — they ask players for screenshots. i want them to see a player's account, recent sessions, purchases, currency history, and be able to grant a compensation reward, all audited, and with no way to see another player's data by accident", "purpose": "planning", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "player lookup by id, email or receipt, returning the account summary with the recent activity", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "player detail screen for support: account, device, recent sessions, purchases, and the currency ledger, dense and scannable", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "the player search matches on a partial email so support sees a list of other players", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.4, "slice": "core", "lang": "en"}
{"prompt": "compensation grants with a reason code, an amount cap per agent, and a full audit entry", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "an agent granted 40,000 gems instead of 400 and there's no cap or confirmation", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "boundary", "lang": "en"}
{"prompt": "review the support tool's authorization — what can an agent do that they shouldn't, and is any of it only hidden in the ui", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "support playbook for the top ten player issues, with the exact tool actions for each and the escalation path", "purpose": "writing", "secondary": null, "mixed": false, "difficulty": 0.5, "slice": "core", "lang": "en"}
{"prompt": "purchase verification against the store receipts so support can confirm a claimed purchase actually happened", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "receipt verification fails for purchases older than 90 days and support treats that as 'no purchase'", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "purchase history view with the store's status, our fulfillment status, and the two reconciled", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "the fulfillment status shows 'pending' for everything older than a week", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.3, "slice": "core", "lang": "en"}
{"prompt": "plan the account recovery flow for players who lost their device and never linked an account, which is 30% of our players", "purpose": "planning", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "account linking prompt in the game that's persuasive without being annoying, shown at a natural moment", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "explain how a guest account's progress is identified today, and what happens if two devices claim the same guest id", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "account recovery documentation for support: what evidence we accept, the risks, and what we never do", "purpose": "writing", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "guest to linked account migration that merges progress deterministically, with the conflict rules decided rather than arbitrary", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.9, "slice": "core", "lang": "en"}
{"prompt": "linking an account wipes the guest progress for some players and keeps it for others", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "our progress merge logic exists in the client and the server with different precedence, so which one wins depends on who runs it. one implementation server side", "purpose": "refactor", "secondary": null, "mixed": false, "difficulty": 0.9, "slice": "core", "lang": "en"}
{"prompt": "keep going", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.3, "slice": "vague-eval", "lang": "en"}
{"prompt": "pasting the crash breakdown from the last release, unity 6, and i genuinely don't know where to start:\n\nVersion 3.9.0 — 4.1% crash rate (was 0.8% on 3.8.2)\n\n  #1  41%  NullReferenceException  InventoryView.OnItemSelected (InventoryView.cs:212)\n  #2  22%  UnityEngine.UnityException  \"CompareBaseObjectsInternal can only be called from the main thread\"\n                                       AssetLoader.OnDownloadComplete (AssetLoader.cs:88)\n  #3  14%  OutOfMemoryException      TextureCache.Insert (TextureCache.cs:141)\n  #4   9%  IndexOutOfRangeException  RewardTrack.GetTier (RewardTrack.cs:64)\n  #5   6%  Native crash (SIGSEGV)    libunity.so — no symbols\n\nDevice skew: #3 is 94% devices with <=3GB RAM. #2 is spread evenly.\n3.9.0 added the addressables migration and the new reward track.", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "pasted-context", "lang": "en"}
{"prompt": "the addressables download callback marshals back to the main thread before touching any unity object", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "texture cache with a memory budget derived from the device's available ram rather than a constant", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "the texture cache budget is 512mb on every device", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.3, "slice": "boundary", "lang": "en"}
{"prompt": "RewardTrack.GetTier indexes past the end when a player's progress exceeds the configured tiers", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.4, "slice": "core", "lang": "en"}
{"prompt": "review the addressables migration for other places we assume a callback is on the main thread", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "write the release retro for 3.9.0 — what got through, why our testing missed it, and what we're changing", "purpose": "writing", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "plan the low-memory device strategy — a quality tier we detect and apply, and what we drop at each level", "purpose": "planning", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "device tier detection with asset variants per tier, and the tier overridable in settings", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "graphics settings screen with the detected tier shown and an explanation of what each option changes", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "our asset loading has an addressables path and a legacy Resources path, both live, and some assets load twice. finish the migration, same assets available", "purpose": "refactor", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "memory grows across scene loads and never comes back on android", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.9, "slice": "core", "lang": "en"}
{"prompt": "explain what our scene unloading actually releases, and what we're holding references to that keeps assets alive", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "asset pipeline documentation: the addressables groups, the tier variants, and the build steps that produce them", "purpose": "writing", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "boundary", "lang": "en"}
{"prompt": "plan the pre-release testing that would have caught a 4% crash rate — device farm coverage and a soak test", "purpose": "planning", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "automated soak test on a device farm: 30 minutes of scripted play across the tiers, failing on a crash or a memory ceiling", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "the soak test runs on three flagship devices and none under 6gb ram", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.3, "slice": "boundary", "lang": "en"}
{"prompt": "check whether our crash reporting captures native crashes with symbols for the android builds", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "design the tier system then implement the detection", "purpose": "planning", "secondary": "backendImpl", "mixed": true, "difficulty": 0.8, "slice": "mixed", "lang": "en"}
{"prompt": "next one please", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.3, "slice": "vague-eval", "lang": "en"}
{"prompt": "i'd like the plan for our in-app purchase pipeline. currently the client tells the server 'i bought thing X' and the server believes it. i want proper receipt validation for both stores, server-authoritative fulfillment, handling of refunds and chargebacks, and a story for the purchases that are stuck in limbo right now", "purpose": "planning", "secondary": null, "mixed": false, "difficulty": 0.9, "slice": "core", "lang": "en"}
{"prompt": "server-side receipt validation for both stores with the fulfillment recorded before the client is told it succeeded", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "the client grants the item locally before the server confirms", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.4, "slice": "boundary", "lang": "en"}
{"prompt": "purchase flow ui: a pending state that's honest, a retry for a failed fulfillment, and a support path when it stays stuck", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "some players are charged and never receive the item, about 40 a day", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.9, "slice": "core", "lang": "en"}
{"prompt": "our purchase handling has separate code paths for ios and android with different retry and idempotency behavior. one flow with store-specific adapters, same outcomes per store", "purpose": "refactor", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "review the purchase path for whether a replayed receipt can grant an item twice", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "purchase and fulfillment documentation for the support and finance teams, including what a refund does to a player's inventory", "purpose": "writing", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "boundary", "lang": "en"}
{"prompt": "refund handling — the store notifies us, we revoke the item if possible and flag the account if the balance is already spent", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "refunded players keep the currency and we have no record of the refund at all", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "stuck purchase recovery job that retries pending fulfillments and reports the ones it can't resolve", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "the recovery job's report goes to a log file nobody reads", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.2, "slice": "core", "lang": "en"}
{"prompt": "purchase reconciliation view: store-reported transactions against our fulfillments, with the mismatches listed", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "plan how we handle store price changes and regional pricing without a client update", "purpose": "planning", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "prices in the shop ui are hardcoded per product rather than read from the store", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.3, "slice": "core", "lang": "en"}
{"prompt": "explain how our shop decides which products to show, and whether a product not available in a region is hidden or errors", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "shop screen with prices from the store, localized currency, and unavailable products hidden rather than failing on tap", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "design the purchase pipeline then implement receipt validation for one store", "purpose": "planning", "secondary": "backendImpl", "mixed": true, "difficulty": 0.9, "slice": "mixed", "lang": "en"}
{"prompt": "look at the purchase code and clean up the duplicated store handling", "purpose": "review", "secondary": "refactor", "mixed": true, "difficulty": 0.7, "slice": "mixed", "lang": "en"}
{"prompt": "and stop there", "purpose": "writing", "secondary": null, "mixed": false, "difficulty": 0.3, "slice": "vague-eval", "lang": "en"}