Live instances
942
369 unique live records
Content operations
The live product has a playable bank today. This page is the expansion system: reusable prompt families, templates, media assets, native-language traps, and new sticky exercise mechanics for a 10,000+ question vault.
Live instances
942
369 unique live records
Generated bank
12,200
1,422 review candidates
Target vault
12,200
2,320 prompts per content cycle
Templates
39
24,672 theoretical variants
Families
11
Reusable skill and mechanic groups
Exercise types
23
11 new sticky mechanics
Seed records
378
375 production-ready seeds
Production queue
A ranked operating view for deciding which generated prompts, seed gaps, source-language lanes, and product surfaces should move next.
Grammar family is high priority, 100% materialized, with 116 sampled drafts ready for review.
300
next batch
100%
target coverage
116
review candidates
0
approved seed gap
8%
review sample
Vocabulary family is high priority, 100% materialized, with 116 sampled drafts ready for review.
300
next batch
100%
target coverage
116
review candidates
0
approved seed gap
8%
review sample
Real life family is high priority, 100% materialized, with 172 sampled drafts ready for review.
240
next batch
100%
target coverage
172
review candidates
0
approved seed gap
14%
review sample
Business English family is high priority, 100% materialized, with 172 sampled drafts ready for review.
220
next batch
100%
target coverage
172
review candidates
0
approved seed gap
14%
review sample
Listening family is high priority, 100% materialized, with 150 sampled drafts ready for review.
220
next batch
100%
target coverage
150
review candidates
0
approved seed gap
13%
review sample
Catalog bridge
This bridge answers the operational question: which catalog tests, guided paths, and product hooks can draw fresh questions from the reusable 10k+ vault instead of depending only on the curated live prompt set.
39
tests with vault routes
39
deep mapped tests
8,725
avg drafts per test
12,200
test-surface drafts
Runtime adapters
The vault no longer only reports inventory. Approved seed records have adapters into the current diagnostic and funnel question shapes, so we can migrate one surface at a time instead of rewriting the runner.
375
approved seeds
375
diagnostic-ready
296
funnel-ready
0
needs custom runner
Scale plan
Core grammar control
High-frequency grammar that affects almost every learner response.
1,500
target
2,160
variants
3
templates
8%
review sample
I bought ___ new laptop yesterday. It was on sale.
Vocabulary and collocation depth
Word choice, collocations, idioms, and precise everyday vocabulary.
1,500
target
3,352
variants
4
templates
8%
review sample
Look at the image. The window is ___.
Business writing and tone
Client-safe tone, concise emails, meetings, and professional repair.
1,200
target
1,800
variants
3
templates
14%
review sample
A client asks for an update, but the project is delayed. Write a short reply.
Listening and fast speech
Reductions, implied meaning, and short spoken details.
1,200
target
2,280
variants
3
templates
12%
review sample
Listen to 'didja get it?' What did you hear?
Natural English usage
Native-sounding phrases, register, and choices that feel safe in context.
1,200
target
2,200
variants
3
templates
10%
review sample
Which sentence sounds most natural: I took a photo or I made a photo?
Visual and simulation mechanics
Playable scenes, shopping missions, instruction sequences, sorting, and word crafting.
1,200
target
3,050
variants
5
templates
14%
review sample
Find the receipt, kettle, and suitcase in the hotel lobby.
Media and pragmatics mechanics
Scene inference, swipe judgments, sound traps, jokes, emotional tone, and synonym depth.
1,000
target
3,500
variants
6
templates
14%
review sample
A waiter lowers his voice and offers to bring something else. What is he doing?
Pronunciation clarity
Sound contrasts, rhythm, word stress, and recordable evidence.
900
target
1,560
variants
3
templates
12%
review sample
Did you hear park or bark?
Reading inference
Main idea, implication, evidence, and paraphrase under compact passages.
900
target
2,040
variants
3
templates
12%
review sample
Read a short workplace notice and choose the main purpose.
Real-life dialogue
Travel, service, appointment, and everyday conversation recovery.
900
target
1,560
variants
3
templates
10%
review sample
You need to reschedule an appointment. What is the best first sentence?
Exam readiness
IELTS, TOEFL, TOEIC, Cambridge, and PTE-style task behavior.
700
target
1,170
variants
3
templates
16%
review sample
Write a short email explaining a delayed delivery and proposing a new date.
10k readiness
10k target
Ready: 12,200 planned records
Expansion targets must add up to a database-sized vault, not a page of examples.
Template capacity
24,672 variants
Prompt templates need more theoretical capacity than the first target bank.
Review sampling
12% generated sample
Generated drafts are not treated as production until sampled and promoted.
Seed provenance
33 seed batches
Materialized seeds retain batch, source, reviewer, and update metadata.
Surface reuse
5 generated surfaces
Seeds and generated drafts are tagged for tests, funnels, daily play, arena, and friend challenges.
Source-language lanes
6 generated lanes
Generated records preserve source-language focus independently from exercise type and surface filters.
Pipeline
Live catalog
Playable prompt instances currently used across tests.
942
Canonical records
Unique live question IDs that can be migrated into a materialized vault.
369
Prompt templates
Reusable generators with slots, assets, quality rules, and max variant limits.
39
Generated draft bank
Deterministic prompt records generated from templates, slots, assets, and quality rules.
12,200
Review candidates
Sampled generated records flagged for human review before production promotion.
1,422
Reviewed seed target
Human-reviewed seeds needed before scaled generation feels controlled.
364
Monthly batches
Planned reviewable prompt output per content cycle.
2,320
Target vault
Reusable prompts across tests, funnels, daily challenges, and friend challenges.
12,200
Materialized
reviewed-seed
375
draft
3
New mechanics
Hidden Object Vocabulary
A1-B2, 45 sec
240 generated, 35 review, 1 source lanes, 5 surfaces
Movie Scene Understanding
A2-C1, 60 sec
167 generated, 24 review, 1 source lanes, 4 surfaces
Swiper Natural/Unnatural
A2-C2, 30 sec
167 generated, 24 review, 2 source lanes, 2 surfaces
Instruction Following
A1-C1, 50 sec
240 generated, 34 review, 2 source lanes, 4 surfaces
Shop Simulator
A1-B2, 70 sec
240 generated, 35 review, 1 source lanes, 4 surfaces
Synonyms Finder
A2-C2, 35 sec
542 generated, 53 review, 2 source lanes, 4 surfaces
Pronunciation Sniper
A1-C1, 40 sec
167 generated, 24 review, 2 source lanes, 3 surfaces
Punchline
B1-C2, 45 sec
166 generated, 24 review, 1 source lanes, 3 surfaces
Emotion Context
A2-C1, 45 sec
166 generated, 23 review, 2 source lanes, 2 surfaces
Sorter
A1-C1, 55 sec
240 generated, 34 review, 1 source lanes, 4 surfaces
Crafter
A2-C2, 75 sec
240 generated, 34 review, 3 source lanes, 4 surfaces
Reusable prompts
levels
24,672 variants39 templates reuse this slot family.
I bought ___ new laptop yesterday. It was on sale.
contexts
10,360 variants13 templates reuse this slot family.
I bought ___ new laptop yesterday. It was on sale.
situations
2,280 variants4 templates reuse this slot family.
Rank four ways to ask a client for a missing file.
passageFrames
1,920 variants3 templates reuse this slot family.
Read a short workplace notice and choose the main purpose.
phraseRules
1,800 variants2 templates reuse this slot family.
Which sentence sounds most natural: I took a photo or I made a photo?
wordFamilies
1,600 variants2 templates reuse this slot family.
In this context, does actual mean real or current?
Asset reuse
Translated phrase distractor rules
3,017 generated drafts reuse this asset. Seed base: 54 x 9.
Verb-noun and adjective-noun collocations
2,559 generated drafts reuse this asset. Seed base: 120 x 8.
Workplace email situations
1,834 generated drafts reuse this asset. Seed base: 64 x 6.
Real-life service dialogue turns
1,700 generated drafts reuse this asset. Seed base: 70 x 7.
Fast speech reductions
1,500 generated drafts reuse this asset. Seed base: 86 x 6.
Mini-passage frames
1,133 generated drafts reuse this asset. Seed base: 48 x 9.
Review and provenance
Review rates
12%
generated sample
99%
approved seeds
Approved human seeds
Owner: content-lead
375
Materialized records that can anchor production and generation examples.
Ready for review
Owner: content-review
0
Authored records waiting on human quality review.
Needs media
Owner: media-production
0
Visual or audio mechanics waiting on generated media assets.
Generated sample review
Owner: content-review
1,422
Deterministic draft records sampled by family review rate.
reviewed-expansion-vocabulary-collocation-depth-2026-05
human-authored from reviewed-expansion-anchor-pack; 28/28approved; updated 2026-05-21.
reviewed-expansion-core-grammar-control-2026-05
human-authored from reviewed-expansion-anchor-pack; 26/26approved; updated 2026-05-21.
reviewed-expansion-business-writing-tone-2026-05
human-authored from reviewed-expansion-anchor-pack; 25/25approved; updated 2026-05-21.
Source language
global
1 traps12,200
generated
371
seed records
es
2 traps3,648
generated
357
seed records
ru
2 traps2,580
generated
344
seed records
ar
1 traps1,067
generated
40
seed records
zh
1 traps1,034
generated
149
seed records
pt
1 traps234
generated
68
seed records
Surface
test
23 types12,200 generated drafts, 1,422 review candidates, 363 seed records.
funnel
16 types4,068 generated drafts, 555 review candidates, 245 seed records.
daily
20 types3,050 generated drafts, 720 review candidates, 375 seed records.
arena
19 types2,440 generated drafts, 457 review candidates, 40 seed records.
friend-challenge
5 types200 generated drafts, 29 review candidates, 261 seed records.
Review status
generated-draft
generated
10,778
generated-review-candidate
generated
1,422
approved
seed-review
375
draft
seed-review
3
Coverage
Synonyms finder
Near misses must be meaningful differences in intensity, tone, or collocation.
Hidden object vocabulary
Every target must be visually findable without exposing the answer label before tap.
Swiper natural/unnatural
Each phrase should reveal one defensible naturalness or transfer signal.
Movie scene understanding
The right answer must require scene context, not a single keyword match.
Sorter
Each item must belong to one category in the given context.
Emotion context
The emotional answer must be inferable from wording, punctuation, and relationship context.
Shop simulator
A successful basket must satisfy the stated English constraints and the budget.
Crafter
The prompt must require productive language, not just memorized word order.
Instruction following
Wrong actions must be plausible enough to test attention, not only vocabulary.
Pronunciation sniper
The round must isolate one sound contrast and use familiar vocabulary.
Punchline
The joke mechanism must be explainable without cultural trivia.
Multiple choice with targeted distractors
Every distractor must map to a named misconception or source-language trap.
Gap fill and typed micro-answer
Accepted answers must be normalized and narrow enough to avoid accidental credit.
Listening choice
Audio text should be short, speech-like, and test one listening signal.
Matching pairs
Pairs must be semantically distinct enough to avoid multiple valid matches.
Reading inference
The key must be supported by the passage, not outside knowledge.
Dialogue completion
The correct answer must fit meaning, tone, and relationship context.
Naturalness ranking
Adjacent ranks must have a clear quality difference reviewers can defend.
Short writing response
The task must create observable tone, grammar, and structure evidence.
Read aloud or repeat
Prompts must include stress, rhythm, or sound focus for scoring.
Minimal pair recognition
Pairs must isolate one sound contrast and avoid rare vocabulary.
Sentence correction
The wrong sentence must be realistic, grammatical-looking, and explainable.
Scenario response
The scenario must be concrete enough that a strong answer is obvious.
Reusable records
After a meeting, someone says: Let’s confirm the action items. What do they mean?
Inspect recordAfter a meeting, someone says: Let’s confirm the action items. What do they mean?
Inspect recordIn a workplace mini-scenario, After a meeting, someone says: Let’s confirm the action items. What do they mean?
Inspect recordYou sent a client the wrong version of a file. You need to fix it.
Inspect recordThe client request is unclear. You need one missing detail before starting.
Inspect recordThe client request is unclear. You need one missing detail before starting.
Inspect recordIn a workplace mini-scenario, The client request is unclear. You need one missing detail before starting.
Inspect recordIn a travel situation, The client request is unclear. You need one missing detail before starting.
Inspect recordA client asks why the report is late. Write a short reply that apologizes, gives a concrete next step, and avoids sounding defensive.
Inspect record