CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

create-marketing-graphics

为活动、促销、品牌 campaign 与倡导等传播任务创建、改版、扩展或诊断海报与营销图形,交付以假乱真的成品平面稿(单张海报、系列、社媒多尺寸套件、展示物与户外幅面)和一个能重出图的母版文件夹(HTML/CSS 画板母版、内容插槽、素材层、色与字规范、导出脚本、README)。基底是固定尺寸的分层版面:生图只产背景与主视觉素材,文字全部活字,系列只变内容不变容器。入口三判据同时成立才属本 Skill:输出是定尺寸单帧静态平面文件(PNG/JPG/PDF,不可滚动、不可交互、不分页);一帧承载全部信息、一次扫读完;用途是传播。用于:海报、传单、KV、banner、社媒图、公众号头图与封面、灯箱、易拉宝、展板、广告图、单页邀请函、电商主图、节日贺图、视频封面、微信长图。不用于并须转出:PPT、报告、手册、杂志、学术海报与任何多页物→design-editorial-and-presentations;logo、VI、名片、品牌手册→design-brand-identities;网页、落地页、dashboard 与任何可交互界面→build-polished-visual-frontends;数据图表与重信息图→create-data-visualizations;单张无文字插画→create-illustrations-and-scenes;单个图标或图形元素→design-icons-and-symbols。

70

Quality

88%

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

73%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.

A rigorously engineered workflow document: gated phases, explicit validation and rework loops, concrete templates, commands, and numeric acceptance criteria, backed by a real, well-organized reference bundle with an on-demand reading index. Its weaknesses are verbosity — the same route/generation/CMYK policies are restated across five or more sections and could be consolidated — and deferral of the most executable details (render invocation syntax, prompt skeletons) to references, which leaves the body's core loop one step short of copy-paste ready.

Suggestions

State each route policy once (e.g., in the §2 route table) and cross-reference it by its 代号 from the principles, gates, §6 signals, and §8 checklist — the CMYK/Ghostscript procedure alone is currently given three times, and consolidating these restatements would cut roughly 30% of the body without losing any rule.

Inline a minimal render_file invocation example (the exact call shape used by the 画板探针 gate) and one image_generate prompt 骨架 so the core generate-then-render loop is executable without opening references/implementation-routing.md.

Move the font-selection table (§2 字体选型) into a reference file (e.g., references/typography.md) and keep only the routing rule and the 中文排版底线 list in the body, matching how the poster-examples index is already handled.

DimensionReasoningScore

Conciseness

The body is entirely operational (no explanations of concepts Claude already knows) but restates the same rules many times: the "SVG 是输出格式不是路线 / 生成层必须存在" rule appears in the five principles, the §2 route table ("不存在 manual_svg、decorative_css、placeholder_as_final 路线"), the §3 tool table, the 素材层 gate, and the §6 signal list; the CMYK-via-Ghostscript rule is given three times (第零步, §3 table, §3 close), and the "不生成的理由都不是理由" list is repeated across §2 and §3. This sits between anchors 2 and 3 — the redundancy is deliberate gate enforcement rather than padded filler, so it lands at 3, but a ~30% cut restating each rule once and cross-referencing by 代号 (P1–P7) would fit anchor 4-5.

3 / 5

Actionability

Guidance is mostly executable: copy-paste-ready YAML templates for the decision card, 最小系统合同, REFERENCE-CONTRACT, and KIT-MANIFEST; concrete commands and parameters ("node master/render.mjs", "render_file(scale=印刷 4 / 大幅面与数字 2, motion_mode=static, output_dir=proof/probe)", "96 px/in", "300 dpi / 150 dpi", "potrace / vtracer", npm font packages); and precise numeric criteria (标题 1/4–1/3 画板高, 信息层底部 1/5, 引用面积 ≥ 35%, 每 3 m 约 25 mm 字高). Minor gaps keep it below 5: the actual render_file/Playwright invocation syntax, an image_generate prompt 骨架 example, and the render.mjs/data.json structure are deferred to reference files rather than shown, and no inline HTML canvas snippet demonstrates the five-layer board.

4 / 5

Workflow Clarity

The sequence is explicitly gated and ordered: 入口 gate (§0) → task reading and 档位 translation → 标杆索引 gate → direction/contract/candidates/fonts → 画板探针 gate before any asset → 素材层 gate before the 定调帧 → hero frame with a pixel-check order → series/master rollout → refactor signals → risk-based checks → completion checklist. Feedback loops are explicit throughout ("探针失败就停下修字体或环境", "生成失败…最多两轮再换 prompt 骨架", "P7 与 P1 不成立时重生成素材或回到最小系统合同"), and the six/seven named rework tests plus the §8 checklist match the anchor-5 pattern of validation steps with error-recovery loops.

5 / 5

Progressive Disclosure

References are one level deep, all verified to exist, and well signaled with purpose and read-condition: "骨架与各档位裁剪见[交付骨架](references/deliverable-skeleton.md)", "按需读取[营销传播模式](references/patterns.md)", and the poster-examples INDEX.md table prescribing exactly which images to read per task. What keeps it at 4 rather than 5: the body itself is ~380 lines where some inlined material (the font-selection table, repeated route-policy restatements) clearly belongs in reference files, and several dependencies are cross-skill paths outside this bundle ("$review-against-ai-patterns", "$visual-artifact-design", "../build-polished-visual-frontends/references/assets-and-imagegen.md") that cannot be navigated from the bundle alone.

4 / 5

Total

16

/

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.

An exceptionally strong description: concrete actions and deliverables, exhaustive natural trigger terms including Chinese platform-specific vocabulary, explicit what/when with entry criteria, and explicit boundary redirects to all six sibling skills. Its only weakness is density — at 615 characters it front-loads implementation policy ("生图只产背景与主视觉素材,文字全部活字,系列只变内容不变容器") that duplicates body content — but this is content-bearing detail, not fluff, so it does not drop any dimension below its best-fit anchor.

DimensionReasoningScore

Specificity

The description lists multiple concrete actions and deliverables: "创建、改版、扩展或诊断海报与营销图形" plus finished artwork types ("单张海报、系列、社媒多尺寸套件、展示物与户外幅面") and an enumerated master folder ("HTML/CSS 画板母版、内容插槽、素材层、色与字规范、导出脚本、README"). Coverage is comprehensive with no vague filler; the level-4 anchor ("minor gaps in coverage") fits worse since nothing in the skill's scope is missing.

5 / 5

Completeness

Both "what" and "when" are answered explicitly and concretely: what — create/extend/diagnose posters and marketing graphics delivering finished artwork plus a re-renderable master folder; when — "入口三判据同时成立才属本 Skill" with three explicit criteria, a "用于:" trigger list, and a "不用于并须转出" exclusion list. This matches the anchor-5 example structure (action list + explicit 'Use when' clause) exactly.

5 / 5

Trigger Term Quality

It covers natural user terms exhaustively with synonyms and platform-specific names: "海报、传单、KV、banner、社媒图、公众号头图与封面、灯箱、易拉宝、展板、广告图、单页邀请函、电商主图、节日贺图、视频封面、微信长图", plus file formats ("PNG/JPG/PDF"). These are phrases a user would actually say; no common variation is missing.

5 / 5

Distinctiveness Conflict Risk

The niche is sharply defined by the three entry criteria (fixed-size single-frame static flat file, one-frame scan, communication purpose) and by explicit redirects of six adjacent skills ("PPT、报告…→design-editorial-and-presentations;logo、VI…→design-brand-identities;网页…→build-polished-visual-frontends;数据图表…→create-data-visualizations;单张无文字插画→create-illustrations-and-scenes;图标→design-icons-and-symbols"). Conflict risk with sibling design skills is minimal because the boundaries are stated rather than implied.

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

referenced_paths_exist

Referenced path issues: 5 missing, 2 deeper-than-1-level

Warning

Total

15

/

16

Passed

Repository

EverMind-AI/Raven

Path

plugins-dist/design-engine/raven_design/skills/create-marketing-graphics/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.