{"prompt": "could you put together the design for our tenant onboarding pipeline? spring boot, and the awkward part is that provisioning touches four systems and any of them can fail halfway. i want the doc before code", "purpose": "planning", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "implement the saga for tenant provisioning as specced in docs/design/provisioning.md — compensations for each step, state in postgres", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "boundary", "lang": "en"}
{"prompt": "spring: add a filter that stamps a correlation id onto MDC and propagates it downstream through the feign clients", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.5, "slice": "core", "lang": "en"}
{"prompt": "would you mind building the account switcher in the ios app? avatar row, checkmark on active, add-account row at the bottom, and switching should tear down the old session cleanly", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "the nav title should be inline not large on the detail screens", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.2, "slice": "boundary", "lang": "en"}
{"prompt": "please bump spring boot to 3.4.1 and deal with whatever deprecations show up", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.3, "slice": "core", "lang": "en"}
{"prompt": "the actuator health endpoint is exposed on the public port, move it to the management port", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.2, "slice": "core", "lang": "en"}
{"prompt": "our service classes are annotated with @Transactional at the class level which means read paths open write transactions. tighten it per-method, semantics preserved where it matters", "purpose": "refactor", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "rename the `Beacon` entity to `Device` across the java code, the flyway migrations going forward, and the openapi models", "purpose": "refactor", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "boundary", "lang": "en"}
{"prompt": "requests occasionally take 5+ seconds with nothing in the app logs and the db query log shows 3ms. hikari pool is 20, we're at maybe 8 concurrent. what else could it be", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "why is the widget timeline only refreshing once a day", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "i'd appreciate a read on our transaction boundaries in the order service, specifically whether the outbox write is inside the same tx as the state change", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "is it safe to run two instances of the scheduler? asking because someone scaled the deployment and i want to know what we're in for", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "boundary", "lang": "en"}
{"prompt": "javadoc on the public service interfaces, and please state the transactional expectations of each method since that's the thing callers get wrong", "purpose": "writing", "secondary": null, "mixed": false, "difficulty": 0.4, "slice": "boundary", "lang": "en"}
{"prompt": "write the swiftui migration notes for the team — what we learned moving the settings screens, patterns to copy, things to avoid", "purpose": "writing", "secondary": null, "mixed": false, "difficulty": 0.4, "slice": "core", "lang": "en"}
{"prompt": "argo rollouts with a canary — 10% for 5 minutes, then 50%, auto-abort on the error rate SLO. plan it first, i want to review before anything lands in the cluster", "purpose": "planning", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "boundary", "lang": "en"}
{"prompt": "helm chart for the new worker service, following the pattern in charts/api, with the hpa and pdb", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.5, "slice": "core", "lang": "en"}
{"prompt": "control room screen for the ops team: service tiles with rag status, last deploy time, and a click-through to logs. dense, no wasted space", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "the pdb minAvailable is 100% which blocks node drains, set it to 50%", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.2, "slice": "core", "lang": "en"}
{"prompt": "our ci has 7 nearly identical jobs. collapse into a matrix, same jobs run, same names in the required checks", "purpose": "refactor", "secondary": null, "mixed": false, "difficulty": 0.5, "slice": "boundary", "lang": "en"}
{"prompt": "nodes go NotReady one at a time every few days, kubelet logs mention PLEG. we're on eks 1.31 with bottlerocket", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.9, "slice": "core", "lang": "en"}
{"prompt": "have a look at our network policies and tell me whether the worker namespace can actually reach the database, i think we have overlapping rules", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "document our deployment process from merge to prod, including the manual approval gate and who can click it", "purpose": "writing", "secondary": null, "mixed": false, "difficulty": 0.4, "slice": "core", "lang": "en"}
{"prompt": "figure out what we do about the 40 pending dependabot PRs, i want a policy not a marathon", "purpose": "planning", "secondary": null, "mixed": false, "difficulty": 0.5, "slice": "core", "lang": "en"}
{"prompt": "yeah go ahead", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.3, "slice": "vague-eval", "lang": "en"}
{"prompt": "ok so here's my problem, i've been going back and forth on this for a week. we need per-customer rate limits but our gateway is envoy and the limits live in a postgres table that the control plane owns, and i don't want a hot path db read. do we push config to envoy, run a local rls service, or something else. write up the options with what each one costs us operationally and pick one", "purpose": "planning", "secondary": null, "mixed": false, "difficulty": 0.9, "slice": "core", "lang": "en"}
{"prompt": "build the rate limit service that envoy's rls will call, redis backed, and a control plane endpoint to update limits without a restart", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "usage meter component showing consumption vs limit with a warning state at 80% and a hard state at 100%", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.4, "slice": "core", "lang": "en"}
{"prompt": "limit header says X-Rate-Limit but the docs say X-RateLimit, pick the docs spelling in the code", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.2, "slice": "boundary", "lang": "en"}
{"prompt": "our config structs are read from viper in 20 places with string keys. load once into a typed struct, pass it down, same values", "purpose": "refactor", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "rate limits are rejecting legitimate traffic from one customer whose usage hasn't changed. counters look wrong but i can't tell why", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "explain the difference between the two ways we can be at 'limit exceeded' — the local token bucket and the global counter. which wins", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "write the customer-facing docs for rate limits: the numbers per plan, the headers, and the recommended backoff", "purpose": "writing", "secondary": null, "mixed": false, "difficulty": 0.4, "slice": "boundary", "lang": "en"}
{"prompt": "i keep thinking about how we'd do soft deletes properly across 40 tables without every query needing to remember the filter. row level security? views? a query layer? write it up", "purpose": "planning", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "add the `restore` mutation for soft-deleted projects, plus the cascade rules for their tasks", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.5, "slice": "core", "lang": "en"}
{"prompt": "trash screen: list of deleted items with days remaining, restore button per row, empty trash in the toolbar", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.5, "slice": "core", "lang": "en"}
{"prompt": "deleted items still show in the count on the sidebar", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.3, "slice": "core", "lang": "en"}
{"prompt": "our repository methods each hand-write `WHERE deleted_at IS NULL`. move it into the base query so a new method can't forget, and verify none of the 90 existing methods change behavior", "purpose": "refactor", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "restoring a project brings back tasks that were deleted separately before the project was. shouldn't happen and i don't see where the timestamps get compared", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "quick sanity check — does our purge job actually delete the s3 objects too or only the db rows", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.4, "slice": "core", "lang": "en"}
{"prompt": "help doc explaining the 30 day trash policy to end users, and what admins can do that regular users can't", "purpose": "writing", "secondary": null, "mixed": false, "difficulty": 0.3, "slice": "core", "lang": "en"}
{"prompt": "here's the stack, this is happening in prod maybe 20 times an hour:\n\njava.lang.IllegalStateException: Cannot forward after response has been committed\n\tat org.apache.catalina.core.ApplicationDispatcher.doForward(ApplicationDispatcher.java:315)\n\tat com.acme.atlas.web.ErrorPageFilter.doFilter(ErrorPageFilter.java:97)\n\tat com.acme.atlas.web.StreamingExportController.export(StreamingExportController.java:64)\n\tat java.base/java.lang.reflect.Method.invoke(Method.java:580)\n\tat org.springframework.web.servlet.mvc.method.annotation.ServletInvocableHandlerMethod.invokeAndHandle(ServletInvocableHandlerMethod.java:118)\nCaused by: org.postgresql.util.PSQLException: An I/O error occurred while sending to the backend.\n\tat org.postgresql.core.v3.QueryExecutorImpl.execute(QueryExecutorImpl.java:379)\n\nthe export streams a large result set. i think the client disconnects mid-stream but i don't understand why that turns into a forward attempt", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "pasted-context", "lang": "en"}
{"prompt": "design the api for scheduled reports and then implement the scheduling part, i'll do the rendering", "purpose": "planning", "secondary": "backendImpl", "mixed": true, "difficulty": 0.7, "slice": "mixed", "lang": "en"}
{"prompt": "rename the confusingly-named `process()` methods across the pipeline classes and update the architecture doc that references them", "purpose": "refactor", "secondary": "writing", "mixed": true, "difficulty": 0.5, "slice": "mixed", "lang": "en"}
{"prompt": "sync stuff again", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.5, "slice": "vague-eval", "lang": "en"}
{"prompt": "here's the review request from the ticket, please take a look and tell me what you think, no changes yet:\n\n```go\nfunc (s *Server) HandleUpload(w http.ResponseWriter, r *http.Request) {\n\tf, hdr, err := r.FormFile(\"file\")\n\tif err != nil {\n\t\thttp.Error(w, err.Error(), 400)\n\t\treturn\n\t}\n\tdefer f.Close()\n\n\tname := filepath.Join(s.uploadDir, hdr.Filename)\n\tout, err := os.Create(name)\n\tif err != nil {\n\t\thttp.Error(w, \"cannot write\", 500)\n\t\treturn\n\t}\n\tdefer out.Close()\n\n\tif _, err := io.Copy(out, f); err != nil {\n\t\thttp.Error(w, \"copy failed\", 500)\n\t\treturn\n\t}\n\tjson.NewEncoder(w).Encode(map[string]string{\"path\": name})\n}\n```", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "pasted-context", "lang": "en"}
{"prompt": "roadmap for our internal developer platform. right now teams copy a template repo and diverge immediately", "purpose": "planning", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "scaffolding generator for new services — repo layout, ci, chart, otel wiring, all from one command", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "service catalog page: search, owner, on-call, tier, links out to dashboards. data comes from the yaml files in the platform repo", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "the template's default cpu limit is 100m which is nonsense, make it 1", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.1, "slice": "core", "lang": "en"}
{"prompt": "extract the shared otel setup from all 12 services into a library. same spans, same attributes, same sampling", "purpose": "refactor", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "traces are missing the db spans for one service only. same library version, same init code as far as i can tell", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "what conventions does the service template actually enforce vs merely suggest", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.4, "slice": "core", "lang": "en"}
{"prompt": "docs for the platform cli — the four commands, what each generates, and how to opt out of the defaults", "purpose": "writing", "secondary": null, "mixed": false, "difficulty": 0.4, "slice": "boundary", "lang": "en"}
{"prompt": "propose how we handle secrets. right now some are in sealed-secrets, some in the aws secrets manager, and two are in a helm values file in git", "purpose": "planning", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "external-secrets operator wiring for the api namespace, with the refresh interval and a fallback if the provider is down at startup", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "the secrets in charts/legacy/values.yaml are committed. rotate them out, that's a today problem", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.4, "slice": "core", "lang": "en"}
{"prompt": "settings form with a masked secret field, reveal on click, and a 'last rotated' timestamp under it", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.4, "slice": "core", "lang": "en"}
{"prompt": "our env var names are inconsistent — some ACME_, some APP_, some bare. unify on ACME_ with a compat shim for one release", "purpose": "refactor", "secondary": null, "mixed": false, "difficulty": 0.5, "slice": "boundary", "lang": "en"}
{"prompt": "pods start fine but 30 minutes later start failing auth to the database. token expiry i assume, but we're using iam auth which is supposed to refresh", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "trace where the db password comes from in each of our four environments, i suspect one of them is still reading the committed file", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "runbook for rotating the database credentials without downtime, including the pgbouncer reload step", "purpose": "writing", "secondary": null, "mixed": false, "difficulty": 0.5, "slice": "core", "lang": "en"}
{"prompt": "ちょっと相談。Kubernetes の Job で動かしているバッチを Argo Workflows に移す方針を書いてほしい。依存関係のあるジョブが 20 個ある", "purpose": "planning", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "ja"}
{"prompt": "sonst nichts, nur den Timeout im Ingress von 60s auf 300s hochsetzen für den Export-Endpoint", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.2, "slice": "core", "lang": "de"}
{"prompt": "swiftui: charts view for the weekly summary using Swift Charts, bar marks, and a selection overlay that shows the day's total", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "background refresh via BGAppRefreshTask, budget aware, and it should coalesce with the push-triggered refresh instead of doubling up", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "pull the networking out of the SwiftUI views — they're calling URLSession directly in .task blocks. put it behind a service, same requests, same order", "purpose": "refactor", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "our app icon on the watch is the iphone one scaled and it looks awful", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.2, "slice": "boundary", "lang": "en"}
{"prompt": "list scrolling stutters on iphone 12 when the images load, and only after you scroll past about 40 rows", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "does our keychain access group setup actually let the widget read the token, or is it falling back to something else", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "boundary", "lang": "en"}
{"prompt": "write the privacy nutrition label details for the app store — what we collect, linked or not, and why", "purpose": "writing", "secondary": null, "mixed": false, "difficulty": 0.4, "slice": "core", "lang": "en"}
{"prompt": "plan how we support the watch app talking to the phone when the phone is offline. what's cached, what's queued, what we tell the user", "purpose": "planning", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "live activity for an in-progress delivery, compact and expanded, updated via push", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "the push token registration retries forever if the first attempt fails, cap it", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.3, "slice": "core", "lang": "en"}
{"prompt": "share the model layer between the app, the widget and the watch target. right now each has its own copy of the structs and they drift", "purpose": "refactor", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "notifications arrive on some devices and not others, same account, both opted in. apns says success for all of them", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "read the sync conflict resolution and tell me what happens if the same field is edited on two devices while both are offline", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "the api reference for our push payload format needs writing, including the silent-push variant and the fields the widget reads", "purpose": "writing", "secondary": null, "mixed": false, "difficulty": 0.4, "slice": "boundary", "lang": "en"}
{"prompt": "everything's fine now", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.3, "slice": "vague-eval", "lang": "en"}
{"prompt": "so i've been staring at our billing code and i think we need to step back. proration is computed in three places, credits are applied after tax in one path and before in another, and nobody knows what happens on a mid-cycle downgrade to a cheaper annual plan. write me the model we SHOULD have, and then how we get from here to there without breaking existing invoices", "purpose": "planning", "secondary": null, "mixed": false, "difficulty": 0.9, "slice": "core", "lang": "en"}
{"prompt": "implement proration for mid-cycle plan changes, credit the unused portion, and produce an invoice preview endpoint that shows the math", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "invoice detail page — line items, the proration rows explained in plain language, download pdf, and a 'why this amount' expander", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "tax is rounded with banker's rounding in one file and half-up in another. make it half-up everywhere, that's what finance expects", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.4, "slice": "boundary", "lang": "en"}
{"prompt": "money is a float in six places. move everything to integer cents with a Money type, and prove no invoice total changes on the last 12 months of data", "purpose": "refactor", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "one customer's invoice is off by $0.03 and the sum of the line items doesn't match the total. only theirs", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "is our webhook handler for `invoice.payment_failed` idempotent? we got three dunning emails for one failure", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "boundary", "lang": "en"}
{"prompt": "write the billing docs for customers: cycles, proration, what happens when you cancel, refunds policy. finance will review so be precise", "purpose": "writing", "secondary": null, "mixed": false, "difficulty": 0.5, "slice": "core", "lang": "en"}
{"prompt": "sequence the work for supporting purchase orders and invoicing outside stripe, enterprise deals need it", "purpose": "planning", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "dunning state machine: retry schedule, emails at each step, downgrade to free after 21 days, and everything auditable", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "payment failed banner, persistent, with a fix-payment cta and a dismissible-until-tomorrow option", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.4, "slice": "core", "lang": "en"}
{"prompt": "the trial length constant says 14 but marketing is advertising 30", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.1, "slice": "core", "lang": "en"}
{"prompt": "pull the entitlement checks out of the controllers into a policy object. same allow/deny for every plan, add tests that enumerate the matrix", "purpose": "refactor", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "users on the team plan sometimes see the free plan limits for a few minutes after upgrading", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "how do we decide a subscription is 'active' — there seem to be three different checks and i want to know which is authoritative", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "release notes for the billing revamp, and a separate internal note for support about what changes in the admin tools", "purpose": "writing", "secondary": null, "mixed": false, "difficulty": 0.4, "slice": "core", "lang": "en"}
{"prompt": "plan the seats-based pricing model change and then build the seat counting service", "purpose": "planning", "secondary": "backendImpl", "mixed": true, "difficulty": 0.8, "slice": "mixed", "lang": "en"}
{"prompt": "walk me through the dunning flow and then document it for support", "purpose": "review", "secondary": "writing", "mixed": true, "difficulty": 0.5, "slice": "mixed", "lang": "en"}
{"prompt": "todo list from standup, take the top one:\n\n- [ ] invoice pdf shows the old logo\n- [ ] proration preview off by a day on month boundaries\n- [ ] webhook retries hammer us during their outages, need backoff\n- [ ] admin can't refund partial amounts\n- [ ] the currency symbol is hardcoded to $ in three emails", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.3, "slice": "pasted-context", "lang": "en"}
{"prompt": "que onda con esto", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.4, "slice": "vague-eval", "lang": "es"}
{"prompt": "i want a design for how the game server handles reconnects. player drops mid-match, we hold their slot for 90 seconds, an ai takes over, and their state has to be exactly restored. write it up including what the client sends on rejoin", "purpose": "planning", "secondary": null, "mixed": false, "difficulty": 0.9, "slice": "core", "lang": "en"}
{"prompt": "authoritative movement validation on the server — reject impossible deltas, allow for 150ms of latency, and log the rejections for tuning", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "scoreboard overlay, sorted live, team colors, and the local player's row highlighted and pinned visible", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.5, "slice": "core", "lang": "en"}
{"prompt": "the tick rate constant is 30, we agreed on 60", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.1, "slice": "core", "lang": "en"}
{"prompt": "our entity update loops are three separate passes over the same list. merge them into one pass, same resulting state each frame", "purpose": "refactor", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "players rubber band near the map edges but only on the eu server. same build as us-east", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "look at the interest management code and tell me how much bandwidth per player we're actually sending at 40 players", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "modding documentation: the lua api surface, the sandbox restrictions, and how to package a mod", "purpose": "writing", "secondary": null, "mixed": false, "difficulty": 0.5, "slice": "boundary", "lang": "en"}
{"prompt": "plan the migration of our save format to something versioned. players have saves from four years of builds", "purpose": "planning", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "replay recording that captures inputs not state, with a compact binary format and a version header", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "settings menu with graphics presets, and changing a preset should show which individual settings it changes before you apply", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "mouse sensitivity slider goes to 100 and the top half is unusable, cap it at 20", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.2, "slice": "core", "lang": "en"}
{"prompt": "our ability system is 4000 lines of switch statements. move to data-driven definitions, same abilities behaving the same in the test scenarios", "purpose": "refactor", "secondary": null, "mixed": false, "difficulty": 0.9, "slice": "core", "lang": "en"}
{"prompt": "frame time spikes to 40ms every ~4 seconds on the main menu of all things. the profiler says gc", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "explain how our netcode handles packet loss right now — is there any redundancy or is it pure reliable-ordered over enet", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "patch notes for the balance pass, and the internal doc explaining the reasoning so we remember next time", "purpose": "writing", "secondary": null, "mixed": false, "difficulty": 0.4, "slice": "core", "lang": "en"}
{"prompt": "postgres logical replication to a read replica in another region for analytics, and please plan the initial sync so it doesn't take the primary down", "purpose": "planning", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "boundary", "lang": "en"}
{"prompt": "materialized view for the daily metrics with a concurrent refresh, and a job that refreshes it every 15 min", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.5, "slice": "core", "lang": "en"}
{"prompt": "the analytics page needs a compare-to-previous-period toggle, with the delta shown per metric", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "statement_timeout isn't set on the analytics role, set it to 30s", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.2, "slice": "core", "lang": "en"}
{"prompt": "the 30 report queries share 5 patterns with slight variations. build them from composable pieces, identical sql output — diff the generated queries to prove it", "purpose": "refactor", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "one report returns different numbers depending on whether you run it in the morning or evening. the date filter looks timezone-correct to me", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "read the query in reports/retention.sql and tell me if the cohort definition matches what we say in the docs", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "boundary", "lang": "en"}
{"prompt": "data dictionary for the warehouse — table by table, column meanings, grain, and which ones are safe to join", "purpose": "writing", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "boundary", "lang": "en"}
{"prompt": "outline how we'd let customers write their own sql against their data safely. read only, resource limited, and no way to see another tenant", "purpose": "planning", "secondary": null, "mixed": false, "difficulty": 0.9, "slice": "core", "lang": "en"}
{"prompt": "sql editor with autocomplete from the schema, run button, results grid, and a cost estimate before you run", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "query timeout error shows the raw pg message to users, wrap it", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.3, "slice": "core", "lang": "en"}
{"prompt": "pgbouncer in transaction mode plus our use of prepared statements is a known bad combo — sort out which one we change, then do it", "purpose": "refactor", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "boundary", "lang": "en"}
{"prompt": "explain the deadlock our logs keep reporting between the two update paths on the inventory table:\n\n2026-07-29 11:04:02.881 UTC [4412] ERROR:  deadlock detected\n2026-07-29 11:04:02.881 UTC [4412] DETAIL:  Process 4412 waits for ShareLock on transaction 88192441; blocked by process 4390.\n\tProcess 4390 waits for ShareLock on transaction 88192438; blocked by process 4412.\n\tProcess 4412: UPDATE inventory SET qty = qty - $1 WHERE sku = $2 AND warehouse_id = $3\n\tProcess 4390: UPDATE inventory SET qty = qty + $1 WHERE warehouse_id = $2 AND sku = ANY($3)\n2026-07-29 11:04:02.881 UTC [4412] HINT:  See server log for query details.\n2026-07-29 11:04:02.881 UTC [4412] CONTEXT:  while updating tuple (441,12) in relation \"inventory\"\n\ni understand deadlocks in general, i want to know specifically what ordering these two take", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "pasted-context", "lang": "en"}
{"prompt": "write up the postgres tuning we did and why, otherwise the next person will undo it", "purpose": "writing", "secondary": null, "mixed": false, "difficulty": 0.4, "slice": "core", "lang": "en"}
{"prompt": "replica lag spikes to 40 seconds during our nightly job and the app serves stale data. the job isn't doing anything huge, 200k updates", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "same as before pls", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.3, "slice": "vague-eval", "lang": "en"}
{"prompt": "we're adding a public api and i need the versioning and deprecation policy written down before we ship anything. header vs path, how long we support a version, what counts as breaking", "purpose": "planning", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "api key management: create, scope, rotate, revoke, last-used tracking, and prefix them so leaks are detectable in scanners", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "api keys page: create modal that shows the key once with a copy button, table with last used and scopes, revoke with confirm", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.5, "slice": "core", "lang": "en"}
{"prompt": "the key prefix is 'sk_' which collides with stripe's, make it 'ak_'", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.2, "slice": "core", "lang": "en"}
{"prompt": "our three auth paths (session, api key, service token) each build a different user context shape. one shape, one builder", "purpose": "refactor", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "api keys stop working after exactly 7 days for some customers. we don't have expiry on keys. i'm confused", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "check whether a revoked key can still be used from a warm cache, and for how long", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "boundary", "lang": "en"}
{"prompt": "authentication docs for the api — the three auth methods, when to use each, code samples in curl, js and python", "purpose": "writing", "secondary": null, "mixed": false, "difficulty": 0.5, "slice": "boundary", "lang": "en"}
{"prompt": "design the audit log we'd need to pass soc2, then implement the middleware that captures it", "purpose": "planning", "secondary": "backendImpl", "mixed": true, "difficulty": 0.8, "slice": "mixed", "lang": "en"}
{"prompt": "clean up the naming in the auth package and update the readme diagram that references the old names", "purpose": "refactor", "secondary": "writing", "mixed": true, "difficulty": 0.5, "slice": "mixed", "lang": "en"}
{"prompt": "sso for the admin console, oidc, and it has to work with google workspace and entra without per-customer code", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "login screen redesign: sso first, email/password behind a 'more options' disclosure, and the error copy shouldn't leak whether an account exists", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.5, "slice": "core", "lang": "en"}
{"prompt": "password field has autocomplete=off which fights password managers, remove it", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.1, "slice": "core", "lang": "en"}
{"prompt": "audit the session handling for fixation and make sure we rotate the session id on privilege change. tell me first, don't change anything", "purpose": "review", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "sso login works but users land on a blank page about 1 in 20 times, and refreshing fixes it", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.7, "slice": "core", "lang": "en"}
{"prompt": "admin guide for configuring sso, per idp, with the exact claim mappings we need and screenshots marked todo", "purpose": "writing", "secondary": null, "mixed": false, "difficulty": 0.5, "slice": "core", "lang": "en"}
{"prompt": "map out what mfa support looks like for us — totp first, webauthn later, recovery codes, and the enforcement policy per org", "purpose": "planning", "secondary": null, "mixed": false, "difficulty": 0.8, "slice": "core", "lang": "en"}
{"prompt": "totp enrollment and verification, drift window of one step, and rate limit the verify endpoint hard", "purpose": "backendImpl", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "mfa setup flow ui: qr code, manual key fallback, verify field with autofocus, then the recovery codes screen with a download", "purpose": "frontendImpl", "secondary": null, "mixed": false, "difficulty": 0.5, "slice": "core", "lang": "en"}
{"prompt": "the totp window is 3 steps which is too loose, make it 1", "purpose": "quickFix", "secondary": null, "mixed": false, "difficulty": 0.2, "slice": "core", "lang": "en"}
{"prompt": "our validation errors come back in two different json shapes depending on which layer rejected. unify the shape, keep the same status codes and messages", "purpose": "refactor", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "recovery codes work twice. they should be single use and i see the delete call right there", "purpose": "debugging", "secondary": null, "mixed": false, "difficulty": 0.6, "slice": "core", "lang": "en"}
{"prompt": "so what now", "purpose": "planning", "secondary": null, "mixed": false, "difficulty": 0.3, "slice": "vague-eval", "lang": "en"}