Sticker & Emoticon Set Pre-Submission Checklist

How it works

Sticker Set Pre-Submission QA is for the day your sticker or emoticon set is finished and before you submit it to a store. It checks the images as one product and tells you what to rework, without touching your files.

Phase 1 — Intake. You state the target platform, the count and size you plan to submit, and which sticker is the reference. Unknown numbers are marked "verify against the platform's current guide".

Phase 2 — The checks. Subject consistency across the set, hands or props covering the face, background halo on transparent PNGs, repeated poses and camera shots, caption and expression spread, spec mismatches, and name-collision suspicion.

Phase 3 — Classify. R1 / R2 / R3 risk levels with evidence per sticker. Criteria are external estimates, never the store's official standard, and no approval numbers are quoted.

Phase 4 — Rework list and re-check. A ranked list that fixes single files instead of redoing the set, then a re-check of only what failed. If the agent cannot view images, it works from your description and marks the rest "not assessable".

Sample output

Real, unedited output from running this Skill on a sample problem:

(QA note: no subagent tool was available in this session; this output was produced by the authoring agent executing SKILL.md strictly. First run revealed two flaws - OK marked where nothing was described, and a messy shot count - so SKILL.md got one fix and this is the rerun.)

**Platform:** LINE Stickers (stated by you). **Spec:** count, pixel size, file format, file-size cap not confirmed (you said "I think 16?", size unknown) - marked "verify against the platform's current guide". **Count found vs expected:** 16 listed vs 16 (unconfirmed). **Access:** described, not viewed. Criteria below are external estimates, not the platform's official review standard.

## Per-sticker table
Reference = #1. OK = description raises no issue on that check (not verified by sight). n/a = not addressed by your text. Filenames were not given.

| # | file | consistency | occlusion | halo/edge | shot/pose | caption/text | level | note |
|---|---|---|---|---|---|---|---|---|
| 1 | n/a | ref | OK | OK | front stand | n/a | - | "Clean cut-out" is your statement, not seen |
| 2 | n/a | n/a | OK | n/a | front stand | n/a | - | "hands clasped", face not covered |
| 3 | n/a | n/a | OK | n/a | front stand | n/a | - | |
| 4 | n/a | n/a | OK | n/a | front stand | n/a | - | |
| 5 | n/a | n/a | FAIL | n/a | front stand | n/a | R1 | "arms crossed over the face area" - emotion-carrying area covered |
| 6 | n/a | n/a | OK | FAIL | front stand | n/a | R1 | "white glow visible around the whole body" on a dark background |
| 7 | n/a | n/a | OK | n/a | front stand | n/a | - | |
| 8 | n/a | FAIL | OK | n/a | front stand | n/a | R1 | "ears look longer than usual and a green leaf shape sits on the head" - off-model, stray element |
| 9 | n/a | n/a | FAIL | n/a | front stand | n/a | R1 | "bowl covers the mouth" |
| 10 | n/a | kept | OK | n/a | close-up | n/a | R3 | intentional variation, not flagged |
| 11 | n/a | n/a | OK | n/a | front stand | n/a | - | |
| 12 | n/a | n/a | OK | n/a | front stand | n/a | - | |
| 13 | n/a | n/a | OK | n/a | front stand | n/a | - | |
| 14 | n/a | n/a | OK | n/a | front stand | n/a | R2 | "same as #1 but with a different mouth" - near-duplicate of Hello |
| 15 | n/a | n/a | OK | n/a | front stand | n/a | - | |
| 16 | n/a | n/a | OK | n/a | lying flat | n/a | - | |

Consistency is `n/a` for all but #8 because your text only describes drift for #8; the others are unchecked, not confirmed. Halo is `n/a` for 14 stickers: only #1 (clean) and #6 (glow) were described. Spec (size, format, transparency, file size) cannot be measured from a text manifest.

## Set-level findings
- **Shot/pose repetition (R1, judged from your text):** front standing = #1,2,3,4,5,6,7,8,9,11,12,13,14,15 = 14; close-up = #10 = 1; lying flat = #16 = 1. Total 16. 14 of 16 share one silhouette, well over the "about half" rule of thumb (an estimate, not a requirement). Emotions differ by face and prop, but may read alike at thumbnail size.
- **Consistency drift:** one reported, #8. Others are unchecked, not cleared.
- **Expression spread:** #1 and #14 are near-duplicates. Greeting, thanks, sorry, OK, no, love, laugh, cry, angry, tired, congrats, bye are covered; no gap that this theme clearly implies.
- **Captions:** small, bottom corner, unchecked at thumbnail size: not assessable.
- **Name-collision suspicion:** "Lazy Bun" - suspect: generic words ("lazy", "bun") that many packs likely use. I cannot search stores. Search the exact title and two variants (for example "Lazy Bunny", "LazyBun") in the target store and the web, and record the result yourself.

## Rework list
1. **#8 R1** - longer ears plus stray leaf shape -> redraw or regenerate this one only, then compare with #1 (a regenerated file can drift in a new way).
2. **#6 R1** (and check the rest) - white glow on dark background -> re-run the cut-out/edge cleanup for this file, then put ALL 16 on a dark and a light background; the same step probably produced others.
3. **#5 R1** - arms over the face area -> move the arms so the eyes/mouth stay visible.
4. **#9 R1** - bowl covers the mouth -> lower or shift the bowl, or show the open mouth above it.
5. **Pose repetition R1** - change the silhouette of several stickers (for example sitting, jumping, three-quarter, bust, lying); fix single files, do not redo the set. Start with those whose emotion needs body language (Sleepy #7, Cheer #12, Angry #11).
6. **#14 R2** - make Bye differ from Hello in pose or prop, or drop one if the count allows.
7. **Captions R3** - view the set at about 100 px and confirm the captions are readable.

## Platform rules to verify
- Verify against the platform's current guide: the sticker count per set (you said "I think 16"), image pixel size, file format and transparency rule, per-file size cap, caption/text rules and language, rules on AI-generated or traced material, naming rules, any rule on glow or shadow edges.

## Limits of this check
- No image was viewed, and nothing was measured; every cell comes from your text. "OK" means "not described as a problem", not "verified". Unaddressed checks are n/a. Name collision was not searched.
Send the reworked files or an updated list and I will re-check only the failed items.

Use cases

• Final check of a 16, 24 or 32 sticker set before submitting to LINE, Kakao, Telegram or another store
• Finding the one off-model sticker among many
• Checking transparent PNGs for halo and fringe
• Spotting a set where almost every sticker shares one pose

FAQ

Does it change or upload my images? No. It is read-only, needs no keys, and stores nothing.

Does it know the store's exact rules? No. Review criteria are external estimates, and size and count rules are marked for you to verify in the platform's current guide.

Can it check images it cannot see? It will say so, work only from your descriptions, and mark everything else "not assessable".

How is it different from a concept review? This one checks finished images; a concept review checks the idea before art exists.