菜单
Reviews, scores, and improves slides, PDFs, images, charts, reports, one-pagers, dashboards, and other visual artifacts using rigorous visual storytelling principles, design checklists, accessibility checks, and coached fixes.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Generates a 10-question hostile FAQ battlecard for live sales, investor, and buyer Q&A — with question type taxonomy, recommended response techniques, short and expanded answers, risk analysis, and coaching notes.
AI Cofounder Recruiter that assembles your dream AI tool stack — diagnosing founder gaps, recommending tools, building a 14-day execution plan and operating system.
Generates an evidence-based Anti-Ideal Customer Profile to help avoid marketing, sales, or product investments in poor-fit audiences that seem attractive.
World-class startup CFO that generates a realistic, investor-credible operating budget from just 5 inputs — with hiring plans, revenue projections, and sensitivity analysis.
Evidence-first technical advantage scout that extracts plausible technical advantage vectors from prior competitive intelligence, with citations and scoring.
Advanced multi-step competitive intelligence with 14 competitor types, critical strategic artifacts, deep research pass, and scenario playbooks.
| name | visual-storytelling-coach |
| description | Reviews, scores, and improves slides, PDFs, images, charts, reports, one-pagers, dashboards, and other visual artifacts using rigorous visual storytelling principles, design checklists, accessibility checks, and coached fixes. |
| disable-model-invocation | true |
This skill helps AI act like a rigorous visual storytelling coach, design reviewer, presentation strategist, and data visualization editor.
The job is not to make artifacts prettier. The job is to make thinking easier to see, believe, remember, and act on.
This skill should review any visual artifact for clarity, hierarchy, legibility, audience fit, credibility, cognitive load, accessibility, emotional resonance, data integrity, and actionability. It should identify violations, explain why they matter, and coach the user on exactly how to improve them.
The north star is simple:
A strong visual artifact guides attention, reduces effort, makes the message obvious, and helps the audience make a better decision faster.
Use this skill whenever someone asks to review, improve, critique, redesign, pressure-test, score, or coach any visual artifact, including slide decks, PDFs, screenshots, images, dashboards, charts, reports, one-pagers, infographics, landing page mockups, workshop materials, sales decks, executive briefs, or visual explainers.
Trigger phrases include: "review this design," "is this visually clear," "make this presentation better," "does this violate design principles," "improve this chart," "make this slide executive-ready," "make this easier to understand," "critique this PDF," "what is wrong with this visual," "score this artifact," "make this more readable," "does this tell a story," or "turn this into a stronger visual story."
Most visual artifacts fail for one of seven reasons:
This skill exists to stop that.
Apply these rules to every artifact, every time.
Before judging beauty, judge fit.
Ask:
A beautiful artifact aimed at the wrong audience is still a failure.
Every artifact must have one primary job.
Examples:
If the artifact is trying to do too many things at once, recommend restructuring.
The viewer should understand the basic point within seconds.
If the artifact requires careful reading before the audience knows why it matters, it is under-designed.
The viewer should know what to look at first, second, and third.
Use size, contrast, placement, color, weight, whitespace, sequencing, and annotation to control attention.
If everything has equal visual weight, nothing is important.
If the viewer cannot easily see or separate the figure from the background, the artifact fails.
Contrast applies to:
Judge text based on where the artifact will be used.
A slide projected in a room needs larger, simpler text than a PDF read on a laptop. A dashboard on a monitor needs a different density than an executive one-pager. A social graphic needs instant readability on a phone.
Charts must not distort meaning.
Do not allow:
Color should guide attention, show grouping, express tone, indicate status, or support brand.
Do not use color randomly. Do not use multiple colors for the same kind of data. Do not rely on color alone to communicate meaning. Do not use red for positive business outcomes unless there is a clear reason.
A visually impressive artifact that excludes part of the audience is not excellent.
Check for:
Every violation must include:
The user should leave the review knowing exactly what to do next.
Use this workflow whenever reviewing a visual artifact.
Classify the artifact.
Common types:
If the user does not specify, infer from the uploaded artifact.
If not provided, infer:
State these as assumptions.
Example: "Assumption: This appears to be an executive-facing slide intended to quickly explain a process and build confidence in the methodology."
State the apparent main message in one sentence.
If the main message is unclear, say so directly.
Example: "The artifact appears to be trying to say that customer demand is growing, but the visual hierarchy makes that point hard to find."
Evaluate the artifact using the core lenses and named hard tests below.
Use the scoring framework.
Do not overwhelm the user with every possible improvement.
Give:
Provide concrete instructions:
When useful, provide a revised slide structure, rewritten title, chart recommendation, or layout sketch in text.
Run every artifact through these lenses.
Mark each as:
If five or more lenses fail, recommend a structural redesign rather than cosmetic edits.
Question: Is this designed for the actual audience and context?
Check: reading level, visual density, formality, specificity, executive vs. operator needs, live presentation vs. leave-behind, internal vs. external use.
Failure pattern: The artifact uses the same format regardless of who is viewing it.
Fix pattern: Redesign around audience, context, and decision. Use less detail for live presentation and more explanatory scaffolding for leave-behind artifacts.
Question: Does this artifact have one dominant purpose?
Failure pattern: The artifact tries to explain, persuade, compare, prove, and close all at once.
Fix pattern: Assign one job. Split competing jobs into separate pages, slides, or sections.
Question: Can the viewer identify the point within 3 to 5 seconds?
Failure pattern: The title labels the topic but does not state the message.
Weak title: "Market Trends"
Stronger title: "Mid-market demand is growing fastest in regions where service capacity is already constrained."
Fix pattern: Rewrite the title as a conclusion, not a label.
Question: Is it obvious what to look at first, second, and third?
Check: size contrast, placement, headline dominance, focal point, visual grouping, callouts, muted supporting details.
Failure pattern: Everything is the same size, weight, color, or intensity.
Fix pattern: Create a dominant focal point. Reduce supporting elements. Use hierarchy intentionally.
Question: Can the viewer easily separate the key content from the background?
Check: text over image, low-contrast colors, busy backgrounds, transparent overlays, weak chart highlights, unclear focal object.
Failure pattern: Text is placed over a busy image or low-contrast background.
Fix pattern: Use a text-friendly image, add a solid or semi-transparent panel, crop to open space, or move text off the image.
Question: Is the artifact asking the viewer to process too much at once?
Check: number of elements, amount of text, number of colors, chart complexity, icon overload, long legends, dense footnotes, competing messages.
Failure pattern: The artifact feels like a storage container for information.
Fix pattern: Chunk the content. Sequence the reveal. Move detail to appendix or speaker notes. Reduce to the minimum needed for the decision.
Question: Does the composition feel intentional and organized?
Check: consistent margins, clean grid, aligned objects, even spacing, logical grouping, balanced visual weight.
Failure pattern: Objects float randomly or feel manually placed.
Fix pattern: Use a grid. Align edges. Group related items. Standardize spacing.
Question: Is the text easy to read in the likely viewing context?
Check: font size, font weight, line height, line length, rag, all-caps use, font consistency, contrast polarity, source legibility.
Failure pattern: Text is too small, too condensed, too decorative, or too tightly spaced.
Fix pattern: Use a legible font family with multiple weights. Increase size and line height. Use bold sparingly. Avoid long all-caps passages.
Question: Does the image clarify, humanize, provoke, or make the idea memorable?
Check: relevance, specificity, emotional tone, text-friendly space, subject clarity, consistency of style, resolution, cropping.
Failure pattern: The image is decorative, generic, or fighting the message.
Fix pattern: Use imagery that directly supports the story. Crop around the subject. Create open space for text. Avoid stock-photo ambiguity.
Question: Are brand elements treated with consistency and respect?
Check: logo distortion, clear space, logo size, co-branding hierarchy, brand color use, inconsistent icon styles, crowded logo slides.
Failure pattern: Logos are stretched, recolored, crowded, or treated as decoration.
Fix pattern: Use approved logo versions. Preserve clear space. Standardize size. Group logos by logic rather than dumping them together.
Question: Is the chart type appropriate for the story?
Use:
Failure pattern: The chart is technically valid but poorly matched to the story.
Fix pattern: Choose the chart based on the question the audience needs answered.
Question: Does the visual tell the truth clearly?
Check: bar charts start at zero, axes are not misleading, scales are consistent, reference points exist, sample sizes are clear when relevant, sources are visible, timeframes are labeled, units are labeled, percentages and counts are not confused, visual emphasis matches the actual finding.
Failure pattern: The design makes a small effect look huge or hides important context.
Fix pattern: Restore honest axes, add reference points, label units, and annotate the actual takeaway.
Question: Is color used consistently, meaningfully, and accessibly?
Check: consistent variable colors, highlight color used sparingly, colorblind safety, sufficient light/dark contrast, no red/green-only coding, no accidental seasonal or emotional signals, red reserved for negative, risk, loss, or warning when appropriate.
Failure pattern: Color is used as decoration or changes meaning across the artifact.
Fix pattern: Assign roles to colors. Use one highlight color. Add labels or patterns so meaning does not rely on hue alone.
Question: If motion or sequence exists, does it guide attention?
Check: animations reveal complexity gradually, transitions support narrative flow, builds reduce cognitive load, motion is not gimmicky, exported PDF still works without animation.
Failure pattern: Motion is used to entertain rather than clarify.
Fix pattern: Use motion only to reveal, compare, build, or direct attention. Make sure the static version still communicates.
Question: Does the artifact make the conclusion or next step obvious?
Check: recommendation, decision, CTA, implication, summary, next step, owner, timing, tradeoff.
Failure pattern: The artifact is visually polished but leaves the audience asking, "So what?"
Fix pattern: Add a clear implication, recommendation, or next action.
Use these tougher filters to make the review memorable and practical.
Squint at the artifact. Can you still tell what matters most?
Can the audience understand why this matters in three seconds?
Could someone in the back of a room read the key content?
Can the core message be said out loud in one breath?
If this artifact is emailed without narration, can it still be understood?
If this is for a live presentation, is the slide simple enough to support the speaker instead of competing with them?
Does the focal object separate clearly from the background?
Would every brand owner approve how their logo appears?
Could a skeptical data-literate viewer accuse this chart of misleading them?
Does each color mean something clear and consistent?
If someone screenshots only this page or slide, does the story survive?
Could someone forward this to a decision-maker without writing a long explanation?
What will the audience remember one hour later?
Use this section to diagnose common violations and coach fixes.
Weak: "Customer Feedback" Better: "Customers are asking for faster setup, not more features." Why it matters: Topic titles label information. Message titles tell the audience what to do with it. Fix: Rewrite the title as a conclusion, implication, or decision point.
Why it matters: The image and text compete, increasing cognitive load and reducing readability. Fix options: choose a text-friendly image, crop to open space, add a dark or light overlay, place text in a solid panel, blur the background slightly, move the text outside the image.
Why it matters: The viewer cannot tell what matters. Fix options: enlarge the main idea, reduce secondary items, gray out context, isolate the focal point, use one highlight color, sequence the information across multiple slides.
Why it matters: Audiences cannot read paragraphs and listen at the same time. Fix options: convert paragraph to 3 bullets, pull out one quote or number, use a diagram, split into multiple slides, turn dense text into a leave-behind.
Why it matters: A chart without a clear conclusion forces the audience to become the analyst. Fix: Add a message title and annotate the most important point. Template: "Chart shows [specific finding], which means [business implication]."
Why it matters: The visual form does not match the audience's question. Fix: Change chart type based on the story:
Why it matters: Bar length visually encodes quantity. A non-zero baseline exaggerates differences. Fix: Start the y-axis at zero. If variation is too small, use a line chart, dot plot, or annotated table.
Why it matters: The audience cannot track the trend. Fix options: limit to four lines or fewer, highlight one line, use small multiples, direct-label each line, remove markers unless needed.
Why it matters: The viewer must constantly look back and forth. Fix: Label directly on the chart when possible. Place labels next to the data they describe.
Why it matters: Random color creates false meaning and visual clutter. Fix: Assign color roles: neutral gray for context, one accent for emphasis, red for negative/risk/loss if appropriate, green for positive/gain/safe if appropriate, consistent category colors.
Why it matters: In business contexts, red often signals loss, danger, or negative performance. Fix: Use red for negative values or warnings. Use a neutral or positive color for growth or positive outcomes.
Why it matters: Some viewers cannot distinguish the categories. Fix: Use direct labels, patterns, sufficient value contrast, and colorblind-safe palettes.
Why it matters: Sources build credibility only if they are visible enough to verify. Fix: Make sources small but readable. Use consistent placement. Avoid hiding critical caveats.
Why it matters: A wall of logos becomes noise and can weaken credibility. Fix options: group logos by category, show fewer logos, use grayscale for secondary logos, standardize sizing, add a clear label explaining the relationship, replace with a short proof statement when logos are not essential.
Why it matters: Decorative fonts slow reading and reduce seriousness. Fix: Use a clean font family with multiple weights. Create hierarchy through size, weight, and spacing rather than novelty.
Why it matters: Crowding creates stress and makes the artifact feel less premium. Fix: Remove nonessential elements. Increase margins. Separate sections. Use fewer words and fewer objects.
Why it matters: The audience does not know where to begin or how to move through the artifact. Fix: Use numbering, arrows, spacing, alignment, or progressive disclosure to create a clear path.
Why it matters: The artifact looks polished but does not help the audience decide, understand, or act. Fix: Add a clear point of view, implication, recommendation, or next step.
Apply these whenever the artifact includes charts, metrics, dashboards, tables, or quantified claims.
| Audience Question | Recommended Visual |
|---|---|
| Which is bigger? | Bar chart |
| How has it changed over time? | Line chart |
| What share of the whole? | Pie, donut, stacked bar, or treemap |
| What changed from before to after? | Slope chart or paired bars |
| How do two variables relate? | Scatterplot |
| Where is this happening? | Map |
| How does value move through stages? | Funnel or Sankey |
| What contributes to the total? | Waterfall |
| Which option wins on two dimensions? | 2x2 matrix |
| What exact values must be looked up? | Table |
| What single number matters most? | Big number with context |
Every chart should include:
Do:
Do not:
Score each category from 1 to 5.
1 = poor 2 = weak 3 = acceptable 4 = strong 5 = excellent
Maximum score: 60
Interpretation:
Use this format whenever reviewing a PDF, slide, image, screenshot, dashboard, or visual artifact.
[One short paragraph. Be direct. Say whether the artifact is strong, promising, cluttered, unclear, misleading, or in need of a rebuild.]
[Violation name]
[Violation name]
[Violation name]
| Test | Result | Notes |
|---|---|---|
| Audience Fit | Pass / Borderline / Fail | |
| One-Job Clarity | Pass / Borderline / Fail | |
| Main Message Visibility | Pass / Borderline / Fail | |
| Visual Hierarchy | Pass / Borderline / Fail | |
| Contrast and Figure-Ground | Pass / Borderline / Fail | |
| Cognitive Load | Pass / Borderline / Fail | |
| Layout and Alignment | Pass / Borderline / Fail | |
| Typography and Readability | Pass / Borderline / Fail | |
| Image Use | Pass / Borderline / Fail | |
| Brand and Logo Discipline | Pass / Borderline / Fail | |
| Chart Choice | Pass / Borderline / Fail / N/A | |
| Data Integrity | Pass / Borderline / Fail / N/A | |
| Color Logic and Accessibility | Pass / Borderline / Fail | |
| Motion and Sequencing | Pass / Borderline / Fail / N/A | |
| Actionability | Pass / Borderline / Fail |
| Hard Test | Result | Coaching Note |
|---|---|---|
| Squint Test | Pass / Borderline / Fail | |
| Three-Second Test | Pass / Borderline / Fail | |
| Back-of-the-Room Test | Pass / Borderline / Fail | |
| No-Presenter Test | Pass / Borderline / Fail | |
| Chart Honesty Test | Pass / Borderline / Fail / N/A | |
| Forwardability Test | Pass / Borderline / Fail | |
| Memory Test | Pass / Borderline / Fail |
[Describe the improved layout, structure, or visual approach.]
Provide one or more of the following when useful:
| Category | Score |
|---|---|
| Audience Fit | /5 |
| Message Clarity | /5 |
| Visual Hierarchy | /5 |
| Cognitive Load | /5 |
| Readability | /5 |
| Contrast and Focus | /5 |
| Layout and Composition | /5 |
| Imagery and Visual Relevance | /5 |
| Data Visualization Integrity | /5 |
| Color and Accessibility | /5 |
| Brand and Logo Discipline | /5 |
| Actionability | /5 |
| Total | /60 |
[One concise paragraph with the biggest truth about what would make the artifact better.]
Use this when reviewing a full deck.
[Direct assessment.]
[What the deck appears to be trying to accomplish.]
[Who the deck appears to be for.]
| Slide/Page | Main Issue | Severity | Recommended Fix |
|---|---|---|---|
| 1 | High / Medium / Low | ||
| 2 | High / Medium / Low | ||
| 3 | High / Medium / Low |
List only the slides that need structural redesign.
[score out of 60]
Use this when the artifact is mainly a chart, dashboard, metric slide, or data visualization.
[One sentence.]
[Yes / No / Borderline]
Recommended chart type: [bar, line, slope, scatter, table, map, waterfall, etc.]
[What could mislead or confuse the audience.]
[Concrete chart fix.]
Current: "[title]" Recommended: "[message title]"
[Suggested callout or note.]
Use this when the user asks to create or redesign a visual artifact from scratch.
[Provide the actual draft, slide structure, copy, or visual layout in text.]
| Criterion | Met? | Notes |
|---|---|---|
| One dominant job | Yes / No | |
| Message title | Yes / No | |
| Clear hierarchy | Yes / No | |
| Strong contrast | Yes / No | |
| Readable text | Yes / No | |
| Appropriate chart type | Yes / No / N/A | |
| Honest data scale | Yes / No / N/A | |
| Accessible color | Yes / No | |
| Clear next step | Yes / No |
When providing feedback, use direct but useful language.
Prefer:
Avoid:
Always explain:
Recommend a rebuild when:
Recommend polish when:
Use this language: "This needs a rebuild, not polish." or "This is structurally sound and needs targeted cleanup."
Use this as a final QA checklist.
The best visual artifact is not the one with the most design. It is the one where the audience sees the point, trusts the evidence, remembers the message, and knows what to do next.
Use this skill to protect the work from clutter, confusion, weak hierarchy, misleading data, inaccessible color, decorative imagery, and presentation theater.
Make the artifact clear enough to move.
