doc_id int64 0 2k | entity_type stringclasses 24
values | prompt stringlengths 795 6.9k | output_format stringclasses 2
values | top_level_count stringclasses 9
values | top_level_key stringclasses 24
values | require_wrapper_key bool 2
classes | require_code_block bool 2
classes | require_no_commentary bool 2
classes | json_schema stringlengths 378 3.03k |
|---|---|---|---|---|---|---|---|---|---|
1,000 | test__escaping__support_ticket_batch | Hi, I'm trying to put together some mock data for a retail platform support queue, basically like what you'd see in a customer support digest handed off for the next shift. I want the values to read as if they were pulled straight from that kind of digest, realistic stuff for support queue triage.
Let me explain the s... | yaml | [2, 3] | support_ticket_batch | true | true | true | {"type": "array", "items": {"type": "object", "properties": {"queue_snapshot_title": {"type": "string", "description": "Short title for the support queue snapshot"}, "support_domain": {"type": "string", "enum": ["auth", "billing", "reporting", "integrations", "infrastructure", "mobile_app", "admin_console", "data_expor... |
1,001 | test__clinical_trial | hey so i need some made-up clinical trial reports, the metabolic/endocrine kind — dose-finding studies where you're establishing the optimal dose. write them like a ClinicalTrials.gov results blurb, precise stats language, all numbers as digits. and keep the data internally coherent, like the p-values should match whet... | yaml | [2, 3] | clinical_trial | false | false | false | {"type": "array", "items": {"type": "object", "properties": {"trial_title": {"type": "string", "description": "Short descriptive trial name (e.g., 'KEYNOTE-789 Phase 3 Trial', 'ATLAS-2 Dose-Ranging Study')"}, "total_enrollment": {"type": "integer", "minimum": 10, "maximum": 10000}, "duration_weeks": {"type": "integer",... |
1,002 | test__gpu_review | Generate exactly 1 compact SFF build GPU review focused on content creation.
Populate the fields as though drawn from a GPU benchmark comparison sheet.
This is a compact SFF build review focused on content creation performance.
Specs and benchmark numbers should be internally consistent and realistic for a current-gene... | json | 1 | gpu_review | true | true | true | {"type": "array", "items": {"type": "object", "properties": {"product_name": {"type": "string", "description": "Full product name including brand and model (e.g., NVIDIA GeForce RTX 4070 Ti Super)"}, "price_usd": {"type": "integer", "minimum": 150, "maximum": 10000}, "rating_out_of_10": {"type": "number", "minimum": 4,... |
1,003 | test__event_ticket_booking | Generate 2-3 event ticket bookings for a music festival day pass booked with parking included.
Assign field values consistent with a ticket order facts table.
This is for a music festival day pass booked with parking included.
Face value, fees, and total paid should add up correctly, and seating details should be plaus... | json | [2, 3] | event_ticket_booking | false | false | true | {"type": "array", "items": {"type": "object", "properties": {"event_name": {"type": "string", "description": "Event name only, not including venue or city"}, "ticket_count": {"type": "integer", "minimum": 1, "maximum": 8}, "face_value_total_usd": {"type": "number", "minimum": 20, "maximum": 10000}, "fees_total_usd": {"... |
1,004 | test__escaping__support_ticket_batch | Generate 1-3 support ticket batches for a SaaS customer support backlog.
Set field values as if extracted from a helpdesk review note for the team lead.
This batch is for a SaaS customer support backlog.
Each case body should feel like a real support message. In each case body, use a realistic multiline support ticket ... | yaml | [1, 3] | support_ticket_batch | false | true | false | {"type": "array", "items": {"type": "object", "properties": {"queue_snapshot_title": {"type": "string", "description": "Short title for the support queue snapshot"}, "backlog_priority": {"type": "string", "enum": ["low", "normal", "elevated", "urgent"]}, "support_domain": {"type": "string", "enum": ["auth", "billing", ... |
1,005 | test__event_ticket_booking | Generate exactly 3 event ticket bookings for a professional sports event booked during a presale window.
Assign field values consistent with an event booking confirmation panel.
This is for a professional sports event booked during a presale window.
Face value, fees, and total paid should add up correctly, and seating ... | json | 3 | event_ticket_booking | true | true | false | {"type": "array", "items": {"type": "object", "properties": {"event_name": {"type": "string", "description": "Event name only, not including venue or city"}, "ticket_count": {"type": "integer", "minimum": 1, "maximum": 8}, "face_value_total_usd": {"type": "number", "minimum": 20, "maximum": 10000}, "fees_total_usd": {"... |
1,006 | test__escaping__screenplay_scene | Generate exactly 4 screenplay scenes for a grounded family drama.
Set field values as though drawn from a coverage packet summarizing selected scene pages.
The material is for a grounded family drama.
Each scene text should read like a real screenplay excerpt rather than prose narration. In each scene text, mix sparse ... | yaml | 4 | screenplay_scenes | false | false | false | {"type": "array", "items": {"type": "object", "properties": {"scene_heading": {"type": "string", "description": "Scene heading line such as INT. APARTMENT - NIGHT"}, "metadata": {"type": "object", "properties": {"genre": {"type": "string", "enum": ["drama", "thriller", "comedy", "sci_fi", "horror", "romance", "mystery"... |
1,007 | test__clinical_trial | Okay so I'm trying to put together some made-up clinical trial reports, the kind of thing you'd see written up in rare diseases and gene therapy work, and specifically I want these to be safety-focused trials that have tolerability data in them. The way I want them to read is like a medical journalist wrote a news arti... | json | [2, 3] | clinical_trial | true | true | true | {"type": "array", "items": {"type": "object", "properties": {"trial_title": {"type": "string", "description": "Short descriptive trial name (e.g., 'KEYNOTE-789 Phase 3 Trial', 'ATLAS-2 Dose-Ranging Study')"}, "duration_weeks": {"type": "integer", "minimum": 1, "maximum": 520}, "phase": {"type": "string", "enum": ["phas... |
1,008 | test__gpu_review | hey, i'm putting together a gpu buyer's guide section focused on content creation performance, and i'd like you to write up some detailed technical reviews for me. think of it like filling out a comparison matrix where the specs and benchmark numbers all hang together realistically for a current-gen card. i need betwee... | yaml | [2, 3] | gpu_review | true | true | true | {"type": "array", "items": {"type": "object", "properties": {"product_name": {"type": "string", "description": "Full product name including brand and model (e.g., NVIDIA GeForce RTX 4070 Ti Super)"}, "price_usd": {"type": "integer", "minimum": 150, "maximum": 10000}, "rating_out_of_10": {"type": "number", "minimum": 4,... |
1,009 | test__escaping__customer_email_thread | ok so I need some fake customer email threads, like the back-and-forth you'd get when someone's having a product access problem. Want exactly 3 of these threads total. And the vibe should be like a customer success recap pulled together from a handful of real emails, if that makes sense — about a product access issue t... | yaml | 3 | customer_email_thread | false | true | false | {"type": "array", "items": {"type": "object", "properties": {"manager_looped_in": {"type": "boolean"}, "thread_emails": {"type": "array", "items": {"type": "object", "properties": {"from_name": {"type": "string", "description": "Display name from the email sender line only"}, "body_text": {"type": "string", "descriptio... |
1,010 | test__rental_car_booking | Generate 1-2 rental car bookings for a family vacation booked with additional driver coverage.
Provide field values matching the detail level of a reservation facts panel from a rental checkout flow.
This is for a family vacation booked with additional driver coverage.
Whenever pricing-related fields are present, keep ... | json | [1, 2] | rental_car_booking | true | false | false | {"type": "array", "items": {"type": "object", "properties": {"rental_company": {"type": "string", "description": "Rental car company name only"}, "vehicle_class": {"type": "string", "enum": ["economy", "compact", "midsize", "full_size", "suv", "minivan", "luxury", "pickup"]}, "daily_rate_usd": {"type": "number", "minim... |
1,011 | test__escaping__screenplay_scene | hey, need you to put together exactly 4 screenplay scenes for a grounded family drama. set the values like they're pulled from a writer's room packet with representative screenplay pages, so it reads like real draft-script context, not commentary about a schema.
on the output: i want yaml, and the whole thing wrapped ... | yaml | 4 | screenplay_scenes | true | true | false | {"type": "array", "items": {"type": "object", "properties": {"scene_heading": {"type": "string", "description": "Scene heading line such as INT. APARTMENT - NIGHT"}, "metadata": {"type": "object", "properties": {"genre": {"type": "string", "enum": ["drama", "thriller", "comedy", "sci_fi", "horror", "romance", "mystery"... |
1,012 | test__escaping__repro_steps_batch | hey so i'm trying to put together some reproduction steps for our api key management page, like the kind of thing you'd see in a bug triage memo where someone's written out the repro instructions for a defect. could you whip up a few of those for me?
i'm thinking like 2 to 3 of these batches, somewhere in that range, ... | json | [2, 3] | repro_steps_batch | true | true | false | {"type": "array", "items": {"type": "object", "properties": {"qa_batch_label": {"type": "string", "description": "Brief label for the set of reproduction cases"}, "repro_cases": {"type": "array", "items": {"type": "object", "properties": {"observed_result": {"type": "string", "enum": ["error_banner", "timeout", "no_res... |
1,013 | test__event_ticket_booking | hey so i need some fake concert ticket bookings, like the kind youd see in a mobile ticket wallet summary, premium seating concert. could you do maybe 5 of them? actually no, lets keep it small, exactly 3 should be fine.
each one needs the event name in `event_name`, just the name of the event no venue or city. and `t... | json | 3 | event_ticket_booking | false | false | false | {"type": "array", "items": {"type": "object", "properties": {"event_name": {"type": "string", "description": "Event name only, not including venue or city"}, "ticket_count": {"type": "integer", "minimum": 1, "maximum": 8}, "face_value_total_usd": {"type": "number", "minimum": 20, "maximum": 10000}, "fees_total_usd": {"... |
1,014 | test__gpu_review | Generate 2-3 detailed technical GPU reviews focused on content creation.
Populate the fields as though drawn from a workstation GPU evaluation worksheet.
This is a detailed technical review focused on content creation performance.
Specs and benchmark numbers should be internally consistent and realistic for a current-g... | yaml | [2, 3] | gpu_review | true | true | true | {"type": "array", "items": {"type": "object", "properties": {"product_name": {"type": "string", "description": "Full product name including brand and model (e.g., NVIDIA GeForce RTX 4070 Ti Super)"}, "price_usd": {"type": "integer", "minimum": 150, "maximum": 10000}, "vram_gb": {"type": "integer", "minimum": 4, "maximu... |
1,015 | test__escaping__bug_report_batch | ok so i need to put together some bug report batches for a developer observability platform, two of them exactly, and i want to be precise about the shape here.
format first... honestly i keep going back and forth, part of me wants yaml because it reads cleaner for multiline bodies, but no, let's do JSON, that's what ... | json | 2 | bug_report_batch | false | true | false | {"type": "array", "items": {"type": "object", "properties": {"triage_label": {"type": "string", "description": "Brief label for the triage batch or defect review"}, "release_bucket": {"type": "string", "enum": ["backlog", "next_sprint", "current_sprint", "blocker"]}, "affected_surface": {"type": "string", "enum": ["api... |
1,016 | test__escaping__screenplay_scene | need 2 screenplay scenes for a backstage theater drama, as JSON. give me exactly 2 of them.
each scene reads like a real screenplay excerpt not prose, with parentheticals/beat notes where natural. fields per scene: `scene_heading` as a string (the heading line like INT. APARTMENT - NIGHT), then `scene_text` as a strin... | json | 2 | screenplay_scenes | true | true | true | {"type": "array", "items": {"type": "object", "properties": {"scene_heading": {"type": "string", "description": "Scene heading line such as INT. APARTMENT - NIGHT"}, "metadata": {"type": "object", "properties": {"genre": {"type": "string", "enum": ["drama", "thriller", "comedy", "sci_fi", "horror", "romance", "mystery"... |
1,017 | test__event_ticket_booking | Generate exactly 3 event ticket bookings for a special holiday performance booked with parking included.
Assign field values consistent with an event booking confirmation panel.
This is for a special holiday performance booked with parking included.
Face value, fees, and total paid should add up correctly, and seating ... | json | 3 | event_ticket_booking | true | true | false | {"type": "array", "items": {"type": "object", "properties": {"event_name": {"type": "string", "description": "Event name only, not including venue or city"}, "ticket_count": {"type": "integer", "minimum": 1, "maximum": 8}, "fees_total_usd": {"type": "number", "minimum": 0, "maximum": 1500}, "total_paid_usd": {"type": "... |
1,018 | test__escaping__short_story_chapter | Generate exactly 3 short story chapters from a social satire set at a formal event.
Fill the fields as though drawn from a manuscript review set quoting chapter excerpts.
The material is from a social satire set at a formal event.
Each chapter text should read like a polished story excerpt, not outline notes. In each c... | yaml | 3 | short_story_chapters | true | true | false | {"type": "array", "items": {"type": "object", "properties": {"metadata": {"type": "object", "properties": {"perspective": {"type": "string", "enum": ["first_person", "close_third", "omniscient", "alternating"]}, "section_role": {"type": "string", "enum": ["opening", "turning_point", "confrontation", "aftermath", "flash... |
1,019 | test__clinical_trial | okay so i need a couple of clinical trial reports, the rare diseases and gene therapy kind, and specifically the gloomy ones — trials that failed, didn't hit significance. write the field values like a medical journalist covering the results, precise stats language, all numbers as digits. and keep it internally coheren... | json | 2 | clinical_trial | true | true | true | {"type": "array", "items": {"type": "object", "properties": {"trial_title": {"type": "string", "description": "Short descriptive trial name (e.g., 'KEYNOTE-789 Phase 3 Trial', 'ATLAS-2 Dose-Ranging Study')"}, "total_enrollment": {"type": "integer", "minimum": 10, "maximum": 10000}, "duration_weeks": {"type": "integer",... |
1,020 | test__escaping__bug_report_batch | Generate exactly 1 bug report batch for a CI and deployment tool.
Fill in all fields as if sourced from a developer handoff note based on bug tracker entries.
This batch is for a CI and deployment tool.
Each bug report body should read like a self-contained defect description, not a summary. In the main report text, in... | yaml | 1 | bug_report_batch | true | true | true | {"type": "array", "items": {"type": "object", "properties": {"triage_label": {"type": "string", "description": "Brief label for the triage batch or defect review"}, "release_bucket": {"type": "string", "enum": ["backlog", "next_sprint", "current_sprint", "blocker"]}, "affected_surface": {"type": "string", "enum": ["api... |
1,021 | test__gpu_review | Hey, could you help me put together some compact SFF build GPU reviews focused on machine learning? Like 2 to 3 of them, numbers should be internally consistent and realistic for a current-gen card. Thanks a ton in advance!
So the core thing I need per card is the `product_name` (full thing with brand and model, e.g. ... | yaml | [2, 3] | gpu_review | true | true | false | {"type": "array", "items": {"type": "object", "properties": {"product_name": {"type": "string", "description": "Full product name including brand and model (e.g., NVIDIA GeForce RTX 4070 Ti Super)"}, "price_usd": {"type": "integer", "minimum": 150, "maximum": 10000}, "rating_out_of_10": {"type": "number", "minimum": 4,... |
1,022 | test__escaping__config_snippet_audit | Hi! Could you help me put together some config snippet audits? I'm working on an ops memo that flags risky config blocks in a container startup configuration, and I want a couple of realistic examples to drop in. Want to make sure I get the structure right so it validates, so bear with me while I think it through.
So ... | yaml | 2 | config_snippet_audit | true | false | false | {"type": "array", "items": {"type": "object", "properties": {"audit_focus": {"type": "string", "description": "Short title for the configuration audit or review"}, "config_family": {"type": "string", "enum": ["yaml", "json", "toml", "env", "ini", "xml", "hcl", "properties"]}, "production_facing": {"type": "boolean"}, "... |
1,023 | test__event_ticket_booking | Generate exactly 2 event ticket bookings for a special holiday performance booked for a group outing.
Assign field values consistent with a ticket order facts table.
This is for a special holiday performance booked for a group outing.
Face value, fees, and total paid should add up correctly, and seating details should ... | yaml | 2 | event_ticket_booking | true | true | true | {"type": "array", "items": {"type": "object", "properties": {"event_name": {"type": "string", "description": "Event name only, not including venue or city"}, "ticket_count": {"type": "integer", "minimum": 1, "maximum": 8}, "fees_total_usd": {"type": "number", "minimum": 0, "maximum": 1500}, "total_paid_usd": {"type": "... |
1,024 | test__camera_review | hey so i'm hoping you can put together some camera reviews for me, like 2 to 3 of them, written from the angle of a street and travel photographer who really cares about autofocus speed and tracking. set the values like they'd show up in a camera testing notes sheet, and keep the sensor specs, pricing and ratings reali... | json | [2, 3] | camera_review | false | true | false | {"type": "array", "items": {"type": "object", "properties": {"product_name": {"type": "string", "description": "Full camera model name including product line (e.g., 'Alpha a7 IV', 'EOS R6 Mark II', 'Z8'), but not the brand"}, "brand": {"type": "string"}, "price_usd": {"type": "integer", "minimum": 400, "maximum": 10000... |
1,025 | test__escaping__dialogue_sample | Need some dialogue samples for a screenwriting exercise — comedian and a heckler going at it backstage. Want maybe 4 or 5 of them... actually let me keep it tighter, between 3 and 4 samples, inclusive.
For each sample, I want two things. First a `metadata` object holding the workshop context (make it read like real wo... | json | [3, 4] | dialogue_samples | false | true | true | {"type": "array", "items": {"type": "object", "properties": {"metadata": {"type": "object", "properties": {"speaker_labels": {"type": "string", "description": "Speaker labels used in the dialogue, such as MANAGER/EMPLOYEE or DETECTIVE/WITNESS"}, "scene_context": {"type": "string", "enum": ["argument", "confession", "ne... |
1,026 | test__camera_review | Generate exactly 1 camera review from a landscape photographer focusing on image quality and dynamic range.
Set field values as they would appear in a photography gear test log.
This is a review from the perspective of a landscape photographer, focusing on image quality and dynamic range.
Sensor specs, pricing, and rat... | json | 1 | camera_review | false | false | false | {"type": "array", "items": {"type": "object", "properties": {"product_name": {"type": "string", "description": "Full camera model name including product line (e.g., 'Alpha a7 IV', 'EOS R6 Mark II', 'Z8'), but not the brand"}, "brand": {"type": "string"}, "price_usd": {"type": "integer", "minimum": 400, "maximum": 10000... |
1,027 | test__camera_review | Generate 2-3 camera reviews from a landscape photographer focusing on image quality and dynamic range.
Set field values as they would appear in a retailer spec-and-rating summary.
This is a review from the perspective of a landscape photographer, focusing on image quality and dynamic range.
Sensor specs, pricing, and r... | json | [2, 3] | camera_review | true | true | true | {"type": "array", "items": {"type": "object", "properties": {"product_name": {"type": "string", "description": "Full camera model name including product line (e.g., 'Alpha a7 IV', 'EOS R6 Mark II', 'Z8'), but not the brand"}, "price_usd": {"type": "integer", "minimum": 400, "maximum": 10000}, "overall_rating": {"type":... |
1,028 | test__escaping__terminal_session_notes | I need some terminal session notes written up for a disk pressure remediation session — basically debugging summaries turned into prose. Can you generate 2-3 of these note sets?
The whole thing should be YAML, wrapped in a single object with the key `terminal_session_notes` pointing to the array of items. So between 2... | yaml | [2, 3] | terminal_session_notes | true | true | true | {"type": "array", "items": {"type": "object", "properties": {"session_summary": {"type": "string", "description": "Short label for the terminal troubleshooting session"}, "shell_family": {"type": "string", "enum": ["bash", "zsh", "powershell", "cmd", "fish"]}, "privileged_access_used": {"type": "boolean"}, "command_fai... |
1,029 | test__real_estate_listing | Generate 1-2 modern property listings across Southeast Asia.
Fill in field values as they might be found in a broker's property summary table.
This is modern property across Southeast Asia.
Price, square footage, lot size, and room counts should be consistent with each other and with the stated property style and regio... | json | [1, 2] | real_estate_listing | true | true | false | {"type": "array", "items": {"type": "object", "properties": {"listing_price_usd": {"type": "integer", "minimum": 50000, "maximum": 10000000}, "property_type": {"type": "string", "enum": ["single_family", "condo", "townhouse", "duplex", "multi_family", "land", "mobile_home"]}, "listing_status": {"type": "string", "enum"... |
1,030 | test__clinical_trial | Generate exactly 3 clinical trial reports in rare diseases and gene therapy describing successful trial meeting its primary endpoint.
Generate field values in the register of a medical conference presentation summary.
This describes a trial in rare diseases and gene therapy, specifically successful trial meeting its pr... | json | 3 | clinical_trial | true | false | true | {"type": "array", "items": {"type": "object", "properties": {"trial_title": {"type": "string", "description": "Short descriptive trial name (e.g., 'KEYNOTE-789 Phase 3 Trial', 'ATLAS-2 Dose-Ranging Study')"}, "total_enrollment": {"type": "integer", "minimum": 10, "maximum": 10000}, "duration_weeks": {"type": "integer",... |
1,031 | test__scientific_experiment | hi, so i'm hoping you can help me put together some fake-but-plausible reports on materials science experiments, like the kind you'd run as a quality control assessment? i'd need somewhere between 1 and 3 of them, and i'd really like the numbers to hang together sensibly — the measurements, sample sizes, stats, all tha... | yaml | [1, 3] | scientific_experiment | true | true | false | {"type": "array", "items": {"type": "object", "properties": {"experiment_title": {"type": "string", "description": "Short descriptive title of the experiment, not including authors or dates"}, "sample_size": {"type": "integer", "minimum": 1, "maximum": 5000}, "num_groups": {"type": "integer", "minimum": 1, "maximum": 2... |
1,032 | test__escaping__short_story_chapter | Generate exactly 2 short story chapters from a post-disaster rebuilding story.
Fill the fields as though drawn from a short fiction craft packet focused on dialogue-heavy chapters.
The material is from a post-disaster rebuilding story.
Each chapter text should read like a polished story excerpt, not outline notes. In e... | json | 2 | short_story_chapters | false | false | true | {"type": "array", "items": {"type": "object", "properties": {"metadata": {"type": "object", "properties": {"perspective": {"type": "string", "enum": ["first_person", "close_third", "omniscient", "alternating"]}, "section_role": {"type": "string", "enum": ["opening", "turning_point", "confrontation", "aftermath", "flash... |
1,033 | test__recipe | hi, could you put together exactly 3 ethiopian lunch recipes for me? i need this as yaml, inside a ```yaml code block, with nothing written outside the data, no commentary.
the root should be a single object with the key `recipe`, and its value is the array of the 3 recipes.
for each recipe i need a `name` (string), ... | yaml | 3 | recipe | true | true | true | {"type": "array", "items": {"type": "object", "properties": {"name": {"type": "string"}, "author": {"type": "string"}, "cuisine": {"type": "string"}, "prep_time_minutes": {"type": "integer", "minimum": 5, "maximum": 120}, "total_time_minutes": {"type": "integer", "minimum": 5, "maximum": 300}, "servings": {"type": "int... |
1,034 | test__clinical_trial | OK so I need a few mock clinical trial reports — metabolic + endocrine disorders, specifically dose-finding trials establishing optimal dosing. Write the field values like a med conference presentation summary, precise stats language, all numbers as digits pls. And keep it internally coherent — p-vals should match whet... | yaml | [2, 3] | clinical_trial | true | true | false | {"type": "array", "items": {"type": "object", "properties": {"trial_title": {"type": "string", "description": "Short descriptive trial name (e.g., 'KEYNOTE-789 Phase 3 Trial', 'ATLAS-2 Dose-Ranging Study')"}, "total_enrollment": {"type": "integer", "minimum": 10, "maximum": 10000}, "duration_weeks": {"type": "integer",... |
1,035 | test__recipe | Hey, so I'm putting together some Ethiopian breakfast recipes and I want them in a really structured format I can validate automatically — let me walk you through what I need.
First off, give me 2-3 of them (so between 2 and 3 items, inclusive). And the way I want it shaped: it should all come back as YAML, and at the... | yaml | [2, 3] | recipe | false | true | true | {"type": "array", "items": {"type": "object", "properties": {"name": {"type": "string"}, "cuisine": {"type": "string"}, "prep_time_minutes": {"type": "integer", "minimum": 5, "maximum": 120}, "cook_time_minutes": {"type": "integer", "minimum": 0, "maximum": 180}, "servings": {"type": "integer", "minimum": 1, "maximum":... |
1,036 | test__escaping__support_ticket_batch | Okay, I need some support ticket batches generated for a payments operations support desk — set the values like you're pulling them from an operations handoff note covering open tickets.
How many batches... let me think. Five feels like a lot for a snapshot. Maybe I'll just ask for exactly 2 to keep it tidy. Actually ... | yaml | [1, 3] | support_ticket_batch | true | true | true | {"type": "array", "items": {"type": "object", "properties": {"queue_snapshot_title": {"type": "string", "description": "Short title for the support queue snapshot"}, "backlog_priority": {"type": "string", "enum": ["low", "normal", "elevated", "urgent"]}, "customer_escalated": {"type": "boolean"}, "cases": {"type": "arr... |
1,037 | test__conference_schedule | Generate 1-2 conference schedules for a nonprofit leadership gathering with networking events built in.
Produce field values that read like entries from a registration site agenda table.
This is for a nonprofit leadership gathering with networking events built in.
Session times, speaker names, and room assignments shou... | yaml | [1, 2] | conference_schedule | false | true | true | {"type": "array", "items": {"type": "object", "properties": {"conference_name": {"type": "string", "description": "Conference or summit name only"}, "conference_type": {"type": "string", "enum": ["conference", "summit", "symposium", "expo", "forum", "retreat"]}, "format": {"type": "string", "enum": ["in_person", "virtu... |
1,038 | test__escaping__bug_report_batch | ok so i need a batch of bug reports for a cloud storage sync service — like the kind of thing you'd pull out of a defect triage recap to hand to the engineering team. so make it feel real, qa triage flavored, not a high-level summary. each bug writeup should read as a self-contained defect description with actual error... | yaml | [2, 3] | bug_report_batch | false | true | true | {"type": "array", "items": {"type": "object", "properties": {"triage_label": {"type": "string", "description": "Brief label for the triage batch or defect review"}, "release_bucket": {"type": "string", "enum": ["backlog", "next_sprint", "current_sprint", "blocker"]}, "hotfix_needed": {"type": "boolean"}, "issue_reports... |
1,039 | test__rental_car_booking | Need a few rental car bookings for a last-minute travel change booked with additional driver coverage — 1-2 of them, detail level like the reservation facts panel at rental checkout. Where pricing fields appear, keep the arithmetic consistent and realistic for the vehicle class.
Fields I want, with their constraints:
... | json | [1, 2] | rental_car_booking | false | true | false | {"type": "array", "items": {"type": "object", "properties": {"rental_company": {"type": "string", "description": "Rental car company name only"}, "vehicle_class": {"type": "string", "enum": ["economy", "compact", "midsize", "full_size", "suv", "minivan", "luxury", "pickup"]}, "daily_rate_usd": {"type": "number", "minim... |
1,040 | test__escaping__pr_discussion | Ok so I need some fake PR discussions for a browser extension codebase — basically the kind of review thread you'd get if you exported a pull request conversation out of the repo UI. So it should feel like a real review going back and forth, not like someone wrote an essay. Diff hunks, inline code suggestions, that kin... | json | [1, 2] | pr_discussion | false | true | true | {"type": "array", "items": {"type": "object", "properties": {"discussion_title": {"type": "string", "description": "Short title for the PR discussion or review thread"}, "review_phase": {"type": "string", "enum": ["draft", "active_review", "revisions", "final_check"]}, "ci_failing": {"type": "boolean"}, "conversation_t... |
1,041 | test__escaping__terminal_session_notes | ok so i need some terminal session notes for a migration and rollback session — the kind of thing you'd see in an engineering diagnostics summary, pasted command-and-output blocks, metadata that'd be plausible for incident response / ops handoff. let me think through the structure.
format first. i keep going back and ... | json | [1, 3] | terminal_session_notes | false | false | true | {"type": "array", "items": {"type": "object", "properties": {"session_summary": {"type": "string", "description": "Short label for the terminal troubleshooting session"}, "shell_family": {"type": "string", "enum": ["bash", "zsh", "powershell", "cmd", "fish"]}, "session_fragments": {"type": "array", "items": {"type": "o... |
1,042 | test__escaping__interview_transcript_segment | hey so i need you to make me some interview transcript segments, like from a startup founder interview. want 4 of them, exactly 4.
format-wise i was leaning toward JSON at first but honestly YAML reads nicer for transcript stuff so let's go YAML.
for the overall structure... hmm, could just be a plain list of the 4 s... | yaml | 4 | interview_transcript_segments | true | true | false | {"type": "array", "items": {"type": "object", "properties": {"metadata": {"type": "object", "properties": {"interview_medium": {"type": "string", "enum": ["in_person", "phone", "video_call", "studio_recording", "field_recording"]}, "interviewer_role": {"type": "string", "enum": ["host", "reporter", "researcher", "hirin... |
1,043 | test__job_posting | Generate exactly 3 healthcare contract job postings.
Supply field values reflecting the tone of a job board listing facts panel.
This is a contract position at a healthcare company.
Salary ranges, benefits, and requirements should be credible for the stated industry and seniority level.
Output valid YAML. Do not wrap ... | yaml | 3 | job_posting | false | true | true | {"type": "array", "items": {"type": "object", "properties": {"job_title": {"type": "string", "description": "Job title only, not including company name or level qualifiers like 'Senior' unless part of the formal title"}, "company_name": {"type": "string"}, "salary_min_usd": {"type": "integer", "minimum": 30000, "maximu... |
1,044 | test__escaping__interview_transcript_segment | ok so i'm trying to put together some fake interview transcript segments, like from a witness interview in a newsroom investigation, and i want them to read like an actual back-and-forth, short questions, long rambly answers, not some tidy summary. the metadata should feel like real transcript context too, not notes ab... | yaml | [2, 3] | interview_transcript_segments | false | true | true | {"type": "array", "items": {"type": "object", "properties": {"metadata": {"type": "object", "properties": {"interview_medium": {"type": "string", "enum": ["in_person", "phone", "video_call", "studio_recording", "field_recording"]}, "interviewer_role": {"type": "string", "enum": ["host", "reporter", "researcher", "hirin... |
1,045 | test__rental_car_booking | Generate 2-3 rental car bookings for a budget-conscious traveler booked through a travel site.
Provide field values matching the detail level of a trip-planning worksheet with vehicle details.
This is for a budget-conscious traveler booked through a travel site.
Whenever pricing-related fields are present, keep them ar... | json | [2, 3] | rental_car_booking | false | false | true | {"type": "array", "items": {"type": "object", "properties": {"rental_company": {"type": "string", "description": "Rental car company name only"}, "vehicle_class": {"type": "string", "enum": ["economy", "compact", "midsize", "full_size", "suv", "minivan", "luxury", "pickup"]}, "daily_rate_usd": {"type": "number", "minim... |
1,046 | test__escaping__terminal_session_notes | hey, so I'm trying to put together some terminal session notes for a feature rollout verification session — basically the kind of thing you'd see in an engineering diagnostics summary, with pasted command-and-output blocks. Could you generate a few of these for me? Like 1-3 of them, somewhere in that range, between 1 a... | yaml | [1, 3] | terminal_session_notes | false | true | true | {"type": "array", "items": {"type": "object", "properties": {"session_summary": {"type": "string", "description": "Short label for the terminal troubleshooting session"}, "shell_family": {"type": "string", "enum": ["bash", "zsh", "powershell", "cmd", "fish"]}, "privileged_access_used": {"type": "boolean"}, "session_fra... |
1,047 | test__conference_schedule | Generate exactly 2 conference schedules for a finance industry summit with hybrid virtual access.
Produce field values that read like entries from an attendee schedule planner.
This is for a finance industry summit with hybrid virtual access.
Session times, speaker names, and room assignments should feel like a real ag... | json | 2 | conference_schedule | true | true | true | {"type": "array", "items": {"type": "object", "properties": {"conference_name": {"type": "string", "description": "Conference or summit name only"}, "registration_fee_usd": {"type": "integer", "minimum": 0, "maximum": 3000}, "conference_type": {"type": "string", "enum": ["conference", "summit", "symposium", "expo", "fo... |
1,048 | test__escaping__log_parser_examples | ok so I need some example sets for parsing Kafka consumer logs, like a little batch I can use for a logging pipeline summary thing built out of parser examples. Let me think through what I actually want in each one because I want this exact.
So each example set needs a label first. I was gonna call it `example_title` ... | yaml | [2, 3] | log_parser_examples | true | true | true | {"type": "array", "items": {"type": "object", "properties": {"parser_review_label": {"type": "string", "description": "Short title for the parser example set or review"}, "source_family": {"type": "string", "enum": ["nginx", "application", "jsonl", "syslog", "windows_event", "audit", "kafka", "etl"]}, "parser_samples":... |
1,049 | test__rental_car_booking | hey so i need a rental car booking generated, just 1 (exactly one item), for a family vacation thats booked with a loyalty member discount. can u give it to me as JSON pls.
heres the thing i care about the structure being exact. the root should be a single object with the key `rental_car_booking` and its value is an a... | json | 1 | rental_car_booking | true | true | true | {"type": "array", "items": {"type": "object", "properties": {"rental_company": {"type": "string", "description": "Rental car company name only"}, "vehicle_class": {"type": "string", "enum": ["economy", "compact", "midsize", "full_size", "suv", "minivan", "luxury", "pickup"]}, "daily_rate_usd": {"type": "number", "minim... |
1,050 | test__escaping__repro_steps_batch | Generate exactly 3 reproduction steps batches for a deployment approval screen.
Write field values as though taken from a troubleshooting summary centered on reproduction attempts.
The batch is for a deployment approval screen.
Repro scripts should read like natural blocks of QA notes. In each repro script, read like r... | json | 3 | repro_steps_batch | true | false | false | {"type": "array", "items": {"type": "object", "properties": {"qa_batch_label": {"type": "string", "description": "Brief label for the set of reproduction cases"}, "test_surface": {"type": "string", "enum": ["web_ui", "mobile_app", "api", "admin_console", "reporting", "billing", "auth", "integrations"]}, "repro_cases": ... |
1,051 | test__clinical_trial | Generate exactly 3 clinical trial reports in respiratory and pulmonary describing trial with mixed results and secondary endpoint hits.
Generate field values in the register of a medical conference presentation summary.
This describes a trial in respiratory and pulmonary, specifically trial with mixed results and secon... | json | 3 | clinical_trial | true | false | true | {"type": "array", "items": {"type": "object", "properties": {"trial_title": {"type": "string", "description": "Short descriptive trial name (e.g., 'KEYNOTE-789 Phase 3 Trial', 'ATLAS-2 Dose-Ranging Study')"}, "total_enrollment": {"type": "integer", "minimum": 10, "maximum": 10000}, "duration_weeks": {"type": "integer",... |
1,052 | test__escaping__config_snippet_audit | Hi, I'm trying to put together some config snippet audits for a feature flag rollout — basically a release readiness review written out as prose. Could you generate a few of these for me?
I'm thinking each one should read like a real platform audit / deployment review, with realistic config blocks embedded. For the co... | json | [2, 3] | config_snippet_audit | false | true | false | {"type": "array", "items": {"type": "object", "properties": {"audit_focus": {"type": "string", "description": "Short title for the configuration audit or review"}, "production_facing": {"type": "boolean"}, "rollback_supported": {"type": "boolean"}, "parser_strictness_known": {"type": "boolean"}, "snippet_findings": {"t... |
1,053 | test__escaping__screenplay_scene | Generate exactly 4 screenplay scenes for a political conspiracy drama.
Set field values as though drawn from a writer's room packet with representative screenplay pages.
The material is for a political conspiracy drama.
Each scene text should read like a real screenplay excerpt rather than prose narration. In each scen... | json | 4 | screenplay_scenes | true | true | true | {"type": "array", "items": {"type": "object", "properties": {"metadata": {"type": "object", "properties": {"genre": {"type": "string", "enum": ["drama", "thriller", "comedy", "sci_fi", "horror", "romance", "mystery", "action"]}, "location_type": {"type": "string", "enum": ["interior", "exterior", "mixed"]}, "time_of_da... |
1,054 | test__escaping__bug_report_batch | hey, thanks so much for helping me out with this one. i'm putting together some sample bug report data for a customer analytics dashboard and i need exactly 3 of these batches. i want them to read like they came straight from a developer handoff note based on real bug tracker entries, so the text should feel authentic ... | json | 3 | bug_report_batch | true | false | true | {"type": "array", "items": {"type": "object", "properties": {"triage_label": {"type": "string", "description": "Brief label for the triage batch or defect review"}, "issue_reports": {"type": "array", "items": {"type": "object", "properties": {"filed_by": {"type": "string", "description": "Name of the reporter or QA own... |
1,055 | test__conference_schedule | Generate 1-3 conference schedules for a product and design event with a workshop-heavy agenda.
Produce field values that read like entries from an event program grid.
This is for a product and design event with a workshop-heavy agenda.
Session times, speaker names, and room assignments should feel like a real agenda wi... | yaml | [1, 3] | conference_schedule | false | true | true | {"type": "array", "items": {"type": "object", "properties": {"conference_name": {"type": "string", "description": "Conference or summit name only"}, "registration_fee_usd": {"type": "integer", "minimum": 0, "maximum": 3000}, "conference_type": {"type": "string", "enum": ["conference", "summit", "symposium", "expo", "fo... |
1,056 | test__escaping__bug_report_batch | Okay, I need some bug report batches put together for a payments integration service — basically pretend you're working off a developer handoff note that pulls from bug tracker entries. The whole point is it should feel real for QA triage and engineering review, so the report bodies need to read like actual self-contai... | yaml | [1, 3] | bug_report_batch | false | true | true | {"type": "array", "items": {"type": "object", "properties": {"triage_label": {"type": "string", "description": "Brief label for the triage batch or defect review"}, "release_bucket": {"type": "string", "enum": ["backlog", "next_sprint", "current_sprint", "blocker"]}, "affected_surface": {"type": "string", "enum": ["api... |
1,057 | test__invoice | hey so I need some invoices for home maintenance services, balance due on completion type deal. finance ops summary sheet vibe, line totals/tax/discounts/final balance should all add up right + be plausible.
format... honestly was leaning YAML for a sec since it's cleaner to eyeball, but nah let's do JSON. and how man... | json | 2 | invoice | true | false | false | {"type": "array", "items": {"type": "object", "properties": {"vendor_name": {"type": "string", "description": "Vendor, contractor, or company issuing the invoice"}, "invoice_total_usd": {"type": "number", "minimum": 50, "maximum": 25000}, "subtotal_usd": {"type": "number", "minimum": 40, "maximum": 23000}, "discount_to... |
1,058 | test__escaping__pr_discussion | Need some sample PR discussion data for an internal admin console — 3 of them. Should read like actual repo discussion notes built from PR comments, so the message turns sound like real review conversation, not standalone docs. Throw in the stuff that shows up in review threads, quoted identifiers or string literals in... | yaml | 3 | pr_discussion | false | true | false | {"type": "array", "items": {"type": "object", "properties": {"discussion_title": {"type": "string", "description": "Short title for the PR discussion or review thread"}, "review_phase": {"type": "string", "enum": ["draft", "active_review", "revisions", "final_check"]}, "conversation_turns": {"type": "array", "items": {... |
1,059 | test__clinical_trial | need 2 clinical trial reports, cardiovascular, safety-focused w/ tolerability data. want it exact pls.
write the field values like a journal abstract draft, key findings register. precise stats language, all numbers as digits. keep it coherent - p-values should match whether primary endpoint met, arm sizes should sum ... | yaml | 2 | clinical_trial | false | true | true | {"type": "array", "items": {"type": "object", "properties": {"trial_title": {"type": "string", "description": "Short descriptive trial name (e.g., 'KEYNOTE-789 Phase 3 Trial', 'ATLAS-2 Dose-Ranging Study')"}, "total_enrollment": {"type": "integer", "minimum": 10, "maximum": 10000}, "duration_weeks": {"type": "integer",... |
1,060 | test__escaping__interview_transcript_segment | Need some interview transcript segments — academic research interview, and I want them to actually read like transcript excerpts. Real Q&A back-and-forth, speaker labels, pauses, quoted speech, the messiness of an actual recording, not a tidied-up summary. Treat the field values like they're pulled from a newsroom back... | json | 3 | interview_transcript_segments | false | true | false | {"type": "array", "items": {"type": "object", "properties": {"metadata": {"type": "object", "properties": {"interview_medium": {"type": "string", "enum": ["in_person", "phone", "video_call", "studio_recording", "field_recording"]}, "interviewer_role": {"type": "string", "enum": ["host", "reporter", "researcher", "hirin... |
1,061 | test__invoice | hey so i need some sample invoices generated, the kind for consulting services with an early payment discount. picturing the facts panel you'd see in a billing portal, and i want the numbers to actually hang together — line totals, tax, discounts, final balance all adding up plausibly.
let me think through what each i... | json | 3 | invoice | false | false | true | {"type": "array", "items": {"type": "object", "properties": {"vendor_name": {"type": "string", "description": "Vendor, contractor, or company issuing the invoice"}, "invoice_total_usd": {"type": "number", "minimum": 50, "maximum": 25000}, "subtotal_usd": {"type": "number", "minimum": 40, "maximum": 23000}, "early_payme... |
1,062 | test__escaping__bug_report_batch | Generate 2-3 bug report batches for a data processing pipeline.
Fill in all fields as if sourced from a QA triage summary rewritten as prose.
This batch is for a data processing pipeline.
Each bug report body should read like a self-contained defect description, not a summary. In the main report text, include a Windows... | json | [2, 3] | bug_report_batch | false | true | false | {"type": "array", "items": {"type": "object", "properties": {"triage_label": {"type": "string", "description": "Brief label for the triage batch or defect review"}, "regression_present": {"type": "boolean"}, "hotfix_needed": {"type": "boolean"}, "issue_reports": {"type": "array", "items": {"type": "object", "properties... |
1,063 | test__escaping__dialogue_sample | Hey, I need a few dialogue samples for a workshop — picture dramaturgy notes that quote dialogue-only sample pages. The setup is always two friends covering for a secret. I want them to read like actual spoken exchanges, not narration, with quoted phrases or repeated wording showing up where it feels natural.
Can you ... | json | [3, 4] | dialogue_samples | false | false | false | {"type": "array", "items": {"type": "object", "properties": {"sample_title": {"type": "string", "description": "Short label for the dialogue excerpt"}, "metadata": {"type": "object", "properties": {"speaker_labels": {"type": "string", "description": "Speaker labels used in the dialogue, such as MANAGER/EMPLOYEE or DETE... |
1,064 | test__escaping__interview_transcript_segment | ok so i need some help generating interview transcript segments, like from a witness interview for a newsroom investigation, and i want them to feel real, actual back-and-forth question and answer stuff, not some tidied up summary. and in each transcript bit i'd like one or two follow-up questions that build on somethi... | yaml | 4 | interview_transcript_segments | false | true | false | {"type": "array", "items": {"type": "object", "properties": {"metadata": {"type": "object", "properties": {"interview_medium": {"type": "string", "enum": ["in_person", "phone", "video_call", "studio_recording", "field_recording"]}, "interviewer_role": {"type": "string", "enum": ["host", "reporter", "researcher", "hirin... |
1,065 | test__scientific_experiment | need 1 microbiology experiment report, industry R&D, metrics consistent / plausible like a grant worksheet.
format wise... i keep going to JSON out of habit, but no, give me YAML. final answer YAML.
each item needs: `experiment_title` as a string, short descriptive title, no authors or dates. `sample_size` integer 1 ... | yaml | 1 | scientific_experiment | true | false | false | {"type": "array", "items": {"type": "object", "properties": {"experiment_title": {"type": "string", "description": "Short descriptive title of the experiment, not including authors or dates"}, "sample_size": {"type": "integer", "minimum": 1, "maximum": 5000}, "num_groups": {"type": "integer", "minimum": 1, "maximum": 2... |
1,066 | test__escaping__terminal_session_notes | ok so we had this webhook failure thing and im trying to put together some terminal session notes for the investigation, like an ops memo where we're quoting the actual shell sessions. need a few of these note sets, somewhere btwn 2 and 3 of them. tone should feel like incident response / ops handoff stuff, realistic c... | json | [2, 3] | terminal_session_notes | false | false | true | {"type": "array", "items": {"type": "object", "properties": {"session_summary": {"type": "string", "description": "Short label for the terminal troubleshooting session"}, "remote_host_used": {"type": "boolean"}, "session_fragments": {"type": "array", "items": {"type": "object", "properties": {"command_goal": {"type": "... |
1,067 | test__camera_review | Generate 1-3 camera reviews from a portrait and wedding photographer focusing on travel and portability.
Set field values as they would appear in a camera testing notes sheet.
This is a review from the perspective of a portrait and wedding photographer, focusing on travel and portability.
Sensor specs, pricing, and rat... | yaml | [1, 3] | camera_review | false | true | true | {"type": "array", "items": {"type": "object", "properties": {"product_name": {"type": "string", "description": "Full camera model name including product line (e.g., 'Alpha a7 IV', 'EOS R6 Mark II', 'Z8'), but not the brand"}, "brand": {"type": "string"}, "price_usd": {"type": "integer", "minimum": 400, "maximum": 10000... |
1,068 | test__escaping__support_ticket_batch | Okay, so I'm trying to mock up some support ticket batches for a SaaS customer support backlog — basically pretend they were extracted from an ops handoff note about open tickets. Let me think through what I actually want here.
First question is the format. I keep going back and forth — JSON would be tidy and easy to ... | yaml | [2, 3] | support_ticket_batch | true | false | false | {"type": "array", "items": {"type": "object", "properties": {"queue_snapshot_title": {"type": "string", "description": "Short title for the support queue snapshot"}, "backlog_priority": {"type": "string", "enum": ["low", "normal", "elevated", "urgent"]}, "customer_escalated": {"type": "boolean"}, "spans_weekend": {"typ... |
1,069 | test__escaping__screenplay_scene | Need some courtroom drama scenes — conflicting testimony angle. 3 to 4 of em, settle on between 3 and 4 inclusive. Workshop handout vibe, like field vals pulled from draft excerpts.
Each scene = an object w/ a few keys. Lemme name em first: `scene_heading`, `metadata`, `scene_text`. Was gonna toss in a `characters` li... | yaml | [3, 4] | screenplay_scenes | false | true | false | {"type": "array", "items": {"type": "object", "properties": {"scene_heading": {"type": "string", "description": "Scene heading line such as INT. APARTMENT - NIGHT"}, "metadata": {"type": "object", "properties": {"genre": {"type": "string", "enum": ["drama", "thriller", "comedy", "sci_fi", "horror", "romance", "mystery"... |
1,070 | test__escaping__screenplay_scene | Generate 3-4 screenplay scenes for a road-trip dramedy.
Set field values as though drawn from a screenplay workshop handout built from draft excerpts.
The material is for a road-trip dramedy.
Each scene text should read like a real screenplay excerpt rather than prose narration. In each scene text, include off-screen o... | json | [3, 4] | screenplay_scenes | false | false | true | {"type": "array", "items": {"type": "object", "properties": {"scene_heading": {"type": "string", "description": "Scene heading line such as INT. APARTMENT - NIGHT"}, "metadata": {"type": "object", "properties": {"genre": {"type": "string", "enum": ["drama", "thriller", "comedy", "sci_fi", "horror", "romance", "mystery"... |
1,071 | test__escaping__bug_report_batch | Hey, could you help me put together some bug report batches for a desktop app? I'm thinking 2-3 batches total, and I want them to read like they came straight out of a defect triage recap for the engineering team. Make the text feel real for QA triage and engineering review.
So each batch should have a few things. The... | json | [2, 3] | bug_report_batch | false | false | true | {"type": "array", "items": {"type": "object", "properties": {"triage_label": {"type": "string", "description": "Brief label for the triage batch or defect review"}, "release_bucket": {"type": "string", "enum": ["backlog", "next_sprint", "current_sprint", "blocker"]}, "regression_present": {"type": "boolean"}, "issue_re... |
1,072 | test__event_ticket_booking | Hi! Hoping you can help me put together a quick bit of mock data. I'm building out an outing planner and need exactly 2 event ticket bookings, each one for a music festival day pass booked with resale protection. The values should hang together sensibly — face value, fees, and total paid adding up correctly, and the se... | json | 2 | event_ticket_booking | false | false | true | {"type": "array", "items": {"type": "object", "properties": {"event_name": {"type": "string", "description": "Event name only, not including venue or city"}, "ticket_count": {"type": "integer", "minimum": 1, "maximum": 8}, "face_value_total_usd": {"type": "number", "minimum": 20, "maximum": 10000}, "total_paid_usd": {"... |
1,073 | test__escaping__repro_steps_batch | ok so i need some test data for a search results interface, basically reproduction steps written up like a qa person logging troubleshooting notes. let me think through how i want this laid out.
i'm picturing a few batches of repro cases. maybe like 5 batches? actually no, let me keep it tighter, exactly 3 batches is ... | yaml | 3 | repro_steps_batch | true | true | true | {"type": "array", "items": {"type": "object", "properties": {"qa_batch_label": {"type": "string", "description": "Brief label for the set of reproduction cases"}, "test_surface": {"type": "string", "enum": ["web_ui", "mobile_app", "api", "admin_console", "reporting", "billing", "auth", "integrations"]}, "customer_repor... |
1,074 | test__recipe | hey, can you put together some Japanese brunch recipes for me? thinking 2-3 of them, somewhere in that range. write the values like you'd see in a proper home cooking tutorial, structured recipe notes style, and keep the timing/servings/ingredient counts all realistic and matching each other.
for each recipe I need a ... | yaml | [2, 3] | recipe | false | true | true | {"type": "array", "items": {"type": "object", "properties": {"name": {"type": "string"}, "cook_time_minutes": {"type": "integer", "minimum": 0, "maximum": 180}, "servings": {"type": "integer", "minimum": 1, "maximum": 48}, "difficulty": {"type": "string", "enum": ["easy", "medium", "hard"]}, "is_gluten_free": {"type": ... |
1,075 | test__escaping__support_ticket_batch | ok so I need some support ticket batches for a logistics platform support queue, like 1 to 3 of them (anywhere from 1 to 3 is fine). Set the values like you pulled them off an internal ticket triage summary.
Let me go through what each batch needs. First theres a `queue_snapshot_title`, thats a string, just a short ti... | yaml | [1, 3] | support_ticket_batch | false | true | true | {"type": "array", "items": {"type": "object", "properties": {"queue_snapshot_title": {"type": "string", "description": "Short title for the support queue snapshot"}, "backlog_priority": {"type": "string", "enum": ["low", "normal", "elevated", "urgent"]}, "customer_escalated": {"type": "boolean"}, "spans_weekend": {"typ... |
1,076 | test__escaping__screenplay_scene | ok so im trying to put together some screenplay scenes — like actual screenplay pages, not prose — for a newsroom drama, the whole everyone's-on-deadline kind of tension. I want it to feel like pages pulled out of a writers room packet, so the metadata should read like real draft-script context, not like notes about th... | json | 3 | screenplay_scenes | true | false | false | {"type": "array", "items": {"type": "object", "properties": {"scene_heading": {"type": "string", "description": "Scene heading line such as INT. APARTMENT - NIGHT"}, "metadata": {"type": "object", "properties": {"genre": {"type": "string", "enum": ["drama", "thriller", "comedy", "sci_fi", "horror", "romance", "mystery"... |
1,077 | test__rental_car_booking | okay so i need some fake rental car bookings for a premium itinerary, prepaid rates kind of thing, and i want to nail down the exact fields before you run with it... bear with me because i keep second guessing myself here
first the count. i was thinking maybe 5 to really pad out the worksheet, or honestly even just 2 ... | json | 3 | rental_car_booking | false | true | true | {"type": "array", "items": {"type": "object", "properties": {"rental_company": {"type": "string", "description": "Rental car company name only"}, "vehicle_class": {"type": "string", "enum": ["economy", "compact", "midsize", "full_size", "suv", "minivan", "luxury", "pickup"]}, "daily_rate_usd": {"type": "number", "minim... |
1,078 | test__escaping__bug_report_batch | Generate exactly 1 bug report batch for a subscription billing platform.
Fill in all fields as if sourced from a release-blocker review summarizing issue reports.
This batch is for a subscription billing platform.
Each bug report body should read like a self-contained defect description, not a summary. In the main repo... | json | 1 | bug_report_batch | false | false | false | {"type": "array", "items": {"type": "object", "properties": {"triage_label": {"type": "string", "description": "Brief label for the triage batch or defect review"}, "release_bucket": {"type": "string", "enum": ["backlog", "next_sprint", "current_sprint", "blocker"]}, "affected_surface": {"type": "string", "enum": ["api... |
1,079 | test__invoice | ok so i need some invoices generated, like 2-3 of them, for a one-time project engagement thats on a recurring billing setup. so each one is basically an export row w totals and terms, and the math should actually add up — line totals, tax, discounts, final balance all plausible for that billing context.
let me think ... | yaml | [2, 3] | invoice | false | true | true | {"type": "array", "items": {"type": "object", "properties": {"vendor_name": {"type": "string", "description": "Vendor, contractor, or company issuing the invoice"}, "invoice_total_usd": {"type": "number", "minimum": 50, "maximum": 25000}, "subtotal_usd": {"type": "number", "minimum": 40, "maximum": 23000}, "tax_total_u... |
1,080 | test__event_ticket_booking | Generate 1-2 event ticket bookings for a music festival day pass booked with mobile entry enabled.
Assign field values consistent with a ticket order facts table.
This is for a music festival day pass booked with mobile entry enabled.
Face value, fees, and total paid should add up correctly, and seating details should ... | json | [1, 2] | event_ticket_booking | true | false | false | {"type": "array", "items": {"type": "object", "properties": {"event_name": {"type": "string", "description": "Event name only, not including venue or city"}, "ticket_count": {"type": "integer", "minimum": 1, "maximum": 8}, "face_value_total_usd": {"type": "number", "minimum": 20, "maximum": 10000}, "fees_total_usd": {"... |
1,081 | test__event_ticket_booking | okay, so what i'm after is some sample event ticket booking data — a special holiday performance, premium seating, that kind of thing. mock data for a confirmation panel, basically. i want it to look believable: the face value, the fees, and the total paid should actually add up, and the seating should make sense for t... | yaml | 3 | event_ticket_booking | true | true | false | {"type": "array", "items": {"type": "object", "properties": {"event_name": {"type": "string", "description": "Event name only, not including venue or city"}, "ticket_count": {"type": "integer", "minimum": 1, "maximum": 8}, "fees_total_usd": {"type": "number", "minimum": 0, "maximum": 1500}, "total_paid_usd": {"type": "... |
1,082 | test__escaping__dialogue_sample | Generate 2-4 dialogue samples between rivals forced to collaborate.
Compose the fields in the spirit of a dramaturgy note quoting dialogue-only sample pages.
The exchange is between rivals forced to collaborate.
Each sample text should read like spoken dialogue rather than narration. In each sample text, include one em... | json | [2, 4] | dialogue_samples | true | false | true | {"type": "array", "items": {"type": "object", "properties": {"sample_title": {"type": "string", "description": "Short label for the dialogue excerpt"}, "metadata": {"type": "object", "properties": {"speaker_labels": {"type": "string", "description": "Speaker labels used in the dialogue, such as MANAGER/EMPLOYEE or DETE... |
1,083 | test__escaping__dialogue_sample | Generate exactly 2 dialogue samples between a traveler and a suspicious customs officer.
Compose the fields in the spirit of a coaching handout built from dramatic dialogue samples.
The exchange is between a traveler and a suspicious customs officer.
Each sample text should read like spoken dialogue rather than narrati... | yaml | 2 | dialogue_samples | true | true | true | {"type": "array", "items": {"type": "object", "properties": {"metadata": {"type": "object", "properties": {"speaker_labels": {"type": "string", "description": "Speaker labels used in the dialogue, such as MANAGER/EMPLOYEE or DETECTIVE/WITNESS"}, "scene_context": {"type": "string", "enum": ["argument", "confession", "ne... |
1,084 | test__recipe | Generate exactly 2 Mexican recipes for dessert.
Write field values as they might appear in a home cooking tutorial written as structured recipe notes.
The recipe should be for a Mexican dish for dessert.
Timing, serving sizes, and ingredient counts should be realistic and consistent with each other.
Return JSON. Use `... | json | 2 | recipe | true | true | false | {"type": "array", "items": {"type": "object", "properties": {"name": {"type": "string"}, "author": {"type": "string"}, "cuisine": {"type": "string"}, "prep_time_minutes": {"type": "integer", "minimum": 5, "maximum": 120}, "cook_time_minutes": {"type": "integer", "minimum": 0, "maximum": 180}, "total_time_minutes": {"ty... |
1,085 | test__recipe | I need some Chinese snack recipes generated, written the way a professional culinary guide would phrase a concise cookbook entry. Let me think through the structure before I pin down the schema.
So the top level — I keep going back and forth here. Part of me wants to wrap it in an object with a `recipes` key for neatn... | json | 2 | recipe | false | true | true | {"type": "array", "items": {"type": "object", "properties": {"name": {"type": "string"}, "prep_time_minutes": {"type": "integer", "minimum": 5, "maximum": 120}, "cook_time_minutes": {"type": "integer", "minimum": 0, "maximum": 180}, "servings": {"type": "integer", "minimum": 1, "maximum": 48}, "difficulty": {"type": "s... |
1,086 | test__job_posting | Generate exactly 1 enterprise software senior individual contributor job posting.
Supply field values reflecting the tone of a hiring intake form for recruiters.
This is a senior individual contributor position at an enterprise software company.
Salary ranges, benefits, and requirements should be credible for the state... | json | 1 | job_posting | false | true | true | {"type": "array", "items": {"type": "object", "properties": {"job_title": {"type": "string", "description": "Job title only, not including company name or level qualifiers like 'Senior' unless part of the formal title"}, "company_name": {"type": "string"}, "salary_max_usd": {"type": "integer", "minimum": 40000, "maximu... |
1,087 | test__escaping__dialogue_sample | hey, can you help me generate some dialogue samples? rivals forced to work together, written like dramaturgy notes quoting dialogue-only sample pages. each one should read like actual spoken dialogue, not narration, with rhetorical questions and abrupt topic changes where they feel natural.
let me think about how many... | yaml | [3, 4] | dialogue_samples | false | true | true | {"type": "array", "items": {"type": "object", "properties": {"metadata": {"type": "object", "properties": {"speaker_labels": {"type": "string", "description": "Speaker labels used in the dialogue, such as MANAGER/EMPLOYEE or DETECTIVE/WITNESS"}, "scene_context": {"type": "string", "enum": ["argument", "confession", "ne... |
1,088 | test__invoice | Hey, I need some mock invoices for event-related services on net-30 terms — finance ops summary sheet style, so the numbers should actually add up (line totals, tax, discounts, final balance all plausible and consistent).
Let me think through what I want before I lock it in. Count first: I keep wanting to say 5 so I h... | yaml | 3 | invoice | true | true | true | {"type": "array", "items": {"type": "object", "properties": {"vendor_name": {"type": "string", "description": "Vendor, contractor, or company issuing the invoice"}, "invoice_total_usd": {"type": "number", "minimum": 50, "maximum": 25000}, "subtotal_usd": {"type": "number", "minimum": 40, "maximum": 23000}, "tax_total_u... |
1,089 | test__escaping__repro_steps_batch | Hey, I need some reproduction steps batches for a data export workflow — generate between 2 and 3 of them. Make the field values read like they came from a real test engineering handoff put together from repro notes, so realistic for defect tracking and troubleshooting handoffs. The repro scripts especially should read... | json | [2, 3] | repro_steps_batch | false | false | true | {"type": "array", "items": {"type": "object", "properties": {"qa_batch_label": {"type": "string", "description": "Brief label for the set of reproduction cases"}, "test_surface": {"type": "string", "enum": ["web_ui", "mobile_app", "api", "admin_console", "reporting", "billing", "auth", "integrations"]}, "blocks_release... |
1,090 | test__escaping__terminal_session_notes | I'm putting together some terminal session notes for a service startup investigation — basically the kind of thing you'd drop into an operations memo where you quote the exact shell sessions you ran while figuring out why a service wouldn't come up. I want it to read like an incident-response/ops-handoff artifact, so t... | yaml | 2 | terminal_session_notes | false | true | false | {"type": "array", "items": {"type": "object", "properties": {"session_summary": {"type": "string", "description": "Short label for the terminal troubleshooting session"}, "session_fragments": {"type": "array", "items": {"type": "object", "properties": {"command_goal": {"type": "string", "enum": ["build", "deploy", "deb... |
1,091 | test__escaping__screenplay_scene | Generate exactly 3 screenplay scenes for a political conspiracy drama.
Set field values as though drawn from a screenplay workshop handout built from draft excerpts.
The material is for a political conspiracy drama.
Each scene text should read like a real screenplay excerpt rather than prose narration. In each scene te... | yaml | 3 | screenplay_scenes | false | true | false | {"type": "array", "items": {"type": "object", "properties": {"scene_heading": {"type": "string", "description": "Scene heading line such as INT. APARTMENT - NIGHT"}, "metadata": {"type": "object", "properties": {"genre": {"type": "string", "enum": ["drama", "thriller", "comedy", "sci_fi", "horror", "romance", "mystery"... |
1,092 | test__real_estate_listing | Generate 1-3 luxury property listings across Eastern Europe.
Fill in field values as they might be found in a listing intake form prepared by an agent.
This is luxury property across Eastern Europe.
Price, square footage, lot size, and room counts should be consistent with each other and with the stated property style ... | json | [1, 3] | real_estate_listing | false | false | false | {"type": "array", "items": {"type": "object", "properties": {"listing_price_usd": {"type": "integer", "minimum": 50000, "maximum": 10000000}, "property_type": {"type": "string", "enum": ["single_family", "condo", "townhouse", "duplex", "multi_family", "land", "mobile_home"]}, "bedrooms": {"type": "integer", "minimum": ... |
1,093 | test__escaping__repro_steps_batch | Hey, can you help me put together some reproduction steps for a subscription management page? I need exactly 3 of these repro batches, and they should read like they came out of a bug triage memo — realistic repro notes, the kind you'd hand off during troubleshooting, not just summaries.
So each batch needs a few thin... | json | 3 | repro_steps_batch | true | true | false | {"type": "array", "items": {"type": "object", "properties": {"qa_batch_label": {"type": "string", "description": "Brief label for the set of reproduction cases"}, "test_surface": {"type": "string", "enum": ["web_ui", "mobile_app", "api", "admin_console", "reporting", "billing", "auth", "integrations"]}, "repro_cases": ... |
1,094 | test__invoice | Generate exactly 2 invoices for a one-time project engagement with a recurring billing arrangement.
Compose field values as they would appear in an invoice export row with totals and terms.
This is for a one-time project engagement with a recurring billing arrangement.
Line totals, tax, discounts, and the final balance... | json | 2 | invoice | true | false | true | {"type": "array", "items": {"type": "object", "properties": {"vendor_name": {"type": "string", "description": "Vendor, contractor, or company issuing the invoice"}, "invoice_total_usd": {"type": "number", "minimum": 50, "maximum": 25000}, "subtotal_usd": {"type": "number", "minimum": 40, "maximum": 23000}, "amount_alre... |
1,095 | test__escaping__dialogue_sample | Generate 2-4 dialogue samples between a detective and a reluctant witness.
Compose the fields in the spirit of a dramaturgy note quoting dialogue-only sample pages.
The exchange is between a detective and a reluctant witness.
Each sample text should read like spoken dialogue rather than narration. In each sample text, ... | json | [2, 4] | dialogue_samples | true | true | true | {"type": "array", "items": {"type": "object", "properties": {"metadata": {"type": "object", "properties": {"speaker_labels": {"type": "string", "description": "Speaker labels used in the dialogue, such as MANAGER/EMPLOYEE or DETECTIVE/WITNESS"}, "scene_context": {"type": "string", "enum": ["argument", "confession", "ne... |
1,096 | test__recipe | Generate 2-3 Indian recipes for dinner.
Write field values as they might appear in a concise cookbook entry in a professional culinary guide.
The recipe should be for a Indian dish for dinner.
Timing, serving sizes, and ingredient counts should be realistic and consistent with each other.
Reply with YAML. Use `recipe`... | yaml | [2, 3] | recipe | true | true | false | {"type": "array", "items": {"type": "object", "properties": {"name": {"type": "string"}, "cuisine": {"type": "string"}, "prep_time_minutes": {"type": "integer", "minimum": 5, "maximum": 120}, "cook_time_minutes": {"type": "integer", "minimum": 0, "maximum": 180}, "total_time_minutes": {"type": "integer", "minimum": 5, ... |
1,097 | test__gpu_review | ok so i'm putting together this little spreadsheet/database thing for picking out a budget gpu mostly for video editing, and i'd love it if you could just generate some sample review entries for me, like the kind of stuff you'd see in a pc builder benchmark log. budget-oriented, video editing focused, and please make t... | json | [1, 2] | gpu_review | true | true | false | {"type": "array", "items": {"type": "object", "properties": {"product_name": {"type": "string", "description": "Full product name including brand and model (e.g., NVIDIA GeForce RTX 4070 Ti Super)"}, "price_usd": {"type": "integer", "minimum": 150, "maximum": 10000}, "rating_out_of_10": {"type": "number", "minimum": 4,... |
1,098 | test__escaping__customer_email_thread | okay so i need a couple of customer email threads, all about a broken integration. exactly 2 of them, not 1 not 3. each one reads like an operations note quoting a support email chain basically.
format wise... hmm. part of me wants JSON since thats what i normally work with, but no — give me YAML for this. yeah YAML.
... | yaml | 2 | customer_email_thread | false | true | true | {"type": "array", "items": {"type": "object", "properties": {"thread_emails": {"type": "array", "items": {"type": "object", "properties": {"from_name": {"type": "string", "description": "Display name from the email sender line only"}, "urgency_level": {"type": "string", "enum": ["low", "normal", "high", "urgent"]}, "ne... |
1,099 | test__escaping__short_story_chapter | need some short story chapters from a sci-fi colony story, intimate stakes. write them like real polished excerpts pulled from a writing residency handout, not outline notes. use em-dash interruptions / broken-off lines where natural, and keep the metadata reading like real manuscript context, not notes about the schem... | json | [3, 4] | short_story_chapters | false | false | false | {"type": "array", "items": {"type": "object", "properties": {"chapter_title": {"type": "string", "description": "Short title or editorial label for the chapter excerpt"}, "metadata": {"type": "object", "properties": {"perspective": {"type": "string", "enum": ["first_person", "close_third", "omniscient", "alternating"]}... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.