CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

sprite-forge

角色动态 sprite / 桌宠皮肤 / atlas 状态行生产线:母图锚身份 → Frame A 首帧 → AI 视频生成(人工环节)→ 截帧确定性加工 → atlas row + 三道闸。 Use when: 做桌宠皮肤、给猫补新状态动画行、visible-cafe/F258 猫形象素材、 角色小循环动画(呼吸/走路/睡觉/挥手级)、把静态角色"做成动态的"。 Not for: 动画短片成片(用 anime-forge)、录屏/教程视频(用 video-forge)、 单张静态图/头像(用 image-generation)、UI 图标。 Output: 192×208 atlas row strip + QA 包(contact/GIF/provenance)→ 过三道闸 + operator identity veto 后接 pet.json。

Quality

96%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Passed

No findings from the security scan

SKILL.md
Quality
Evals
Security

Quality

Content

92%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

An exemplary operational skill body: a sequenced pipeline with named roles and validation gates, copy-paste deterministic processing commands, hard-won failure rules that encode knowledge Claude doesn't have, and heavy material correctly pushed to on-demand reference files. The only weakness is minor token overhead from the origin story and repeated incident-date citations.

DimensionReasoningScore

Conciseness

The body is dense and assumes competence — commands are given bare, no concept explanations, and the failure lessons (工艺红线, Common Mistakes) are exactly what Claude couldn't know. Minor trimmable overhead remains: the origin narrative ("> 出生:F229 … operator 连续两次 identity veto 后拉闸") and repeated date citations ("2026-07-08 实录", "2026-06-22~24") are provenance, not instructions. This fits anchor 4 (efficient, minor instances that could be trimmed) better than anchor 5's 'every token earns its place'.

4 / 5

Actionability

The Step-4 chain is copy-paste ready with exact parameters ("ffmpeg -i video.mp4 -vf \"fps=4\" frame_%02d.png", "magick f1.png -resize 192x208 -gravity center -extent 192x208 cell_01.png", "-resize 59x64 preview.png # 必做"), the prompt idiom is given verbatim ("Only differences: ... Everything else IDENTICAL"), and fallbacks for common failure cases are concrete ("B 帧锁不住… → 降级单图 i2v 只用 A 帧", "脏边则重跑 despill"). Specific examples cover the common cases, matching anchor 5.

5 / 5

Workflow Clarity

The 5-step pipeline table sequences who/what/constraints per step, and validation is explicit and layered: "三道闸 + operator identity veto 终审", "每批过了才做下一批(2-2-3 批次制先例)" — a batch operation with per-batch gates, so the missing-validation cap of 3 does not apply. Error-recovery feedback loops are present (fallback to single-image i2v, "抽卡 2-3 败即换姿势写法别硬 roll", "修正成功版按红线 6 回流"). This matches anchor 5's 'explicit validation steps; feedback loops; checklists'.

5 / 5

Progressive Disclosure

The body is a ~70-line overview and explicitly delegates heavy material: "真相源(重材料按需读,不复制进上下文)" lists four one-level-deep references, each labeled with its content (prompt template library, prompt pack, cell spec/gates, pet.json format example). No bundle files exist in this review workspace to verify the external repo paths, but as written the structure matches anchor 5: clear overview, well-signaled one-level references, easy navigation.

5 / 5

Total

19

/

20

Passed

Description

100%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A model description: concrete pipeline stages and an exact output artifact, explicit Use-when triggers phrased the way users would actually talk, an explicit Not-for boundary that routes collisions to sibling skills, and third-person voice throughout. No fluff or over-claims.

DimensionReasoningScore

Specificity

The description lays out the full pipeline concretely ("母图锚身份 → Frame A 首帧 → AI 视频生成(人工环节)→ 截帧确定性加工 → atlas row + 三道闸") and a concrete output spec ("Output: 192×208 atlas row strip + QA 包(contact/GIF/provenance)→ 过三道闸 + operator identity veto 后接 pet.json"). It covers input, process stages, and output comprehensively with no vague filler, matching the 'multiple specific concrete actions; comprehensive coverage' anchor; anchor 4 would require coverage gaps that aren't present.

5 / 5

Completeness

It explicitly answers what (the pipeline and output format), when (an explicit "Use when:" clause with concrete triggers), and additionally when-not ("Not for: 动画短片成片(用 anime-forge)… 单张静态图/头像… UI 图标"). The explicit Use-when clause with concrete trigger phrases matches the anchor-5 example exactly; the guideline cap of 3 for a missing Use-when does not apply.

5 / 5

Trigger Term Quality

Natural user phrasings span the domain with synonyms: "做桌宠皮肤", "给猫补新状态动画行", "角色小循环动画(呼吸/走路/睡觉/挥手级)", "把静态角色'做成动态的'". A user needing this skill would plausibly say one of these verbatim; coverage is comprehensive rather than 'a few natural terms missing' (anchor 4).

5 / 5

Distinctiveness Conflict Risk

A clear niche (looping state-row sprite assets for pet skins) is reinforced by explicit routing away from sibling skills: "Not for: 动画短片成片(用 anime-forge)、录屏/教程视频(用 video-forge)、单张静态图/头像(用 image-generation)、UI 图标". Conflict risk is minimal because wrong-skill triggers are explicitly redirected.

5 / 5

Total

20

/

20

Passed

Validation

93%

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 15 / 16 Passed

Validation for skill structure

CriteriaDescriptionResult

frontmatter_unknown_keys

Unknown frontmatter key(s) found; consider removing or moving to metadata

Warning

Total

15

/

16

Passed

Repository

zts212653/clowder-ai

Path

cat-cafe-skills/sprite-forge/SKILL.md

Reviewed

Table of Contents

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.