authentic-product-representation
Skill194 starsChanged 4 months agoMITView on GitHub
Product visuals must show the real product with real data. Use when building landing pages, hero sections, screenshots or demo panels.
In this repository44 other files it gives its agents.
- algorithmic-color-paletteSkill
- app-shellSkill
- brand-visual-languageSkill
- button-statesSkill
- clone-websiteSkill
- color-mode-and-themeSkill
- component-family-consistencySkill
- coordinated-data-viewsSkill
- data-display-and-selectionSkill
- dembrandtSkill
- domain-expert-configurationSkill
- elevation-and-depthSkill
- extract-designSkill
- form-designSkill
- generate-ui-from-brandSkill
- gestalt-ui-organisationSkill
- global-toolbar-controlsSkill
- information-architectureSkill
- layout-paradigms-and-consistencySkill
- loading-states-and-perceived-performanceSkill
- marketing-vs-product-systemSkill
- micro-interactionsSkill
- modal-and-overlay-patternsSkill
- modular-scale-typographySkill
- motion-and-storytellingSkill
- nielsen-usability-heuristicsSkill
- notifications-and-recoverySkill
- operational-expert-tool-uiSkill
- performance-and-web-vitalsSkill
- real-world-metaphorsSkill
- repeated-component-alignmentSkill
- responsive-paradigmsSkill
- scroll-areasSkill
- semantic-html-and-seoSkill
- sizing-unitsSkill
- status-colors-and-errorsSkill
- sticky-and-fixed-elementsSkill
- surface-separationSkill
- tab-navigationSkill
- ui-context-and-scopeSkill
- ui-densitySkill
- user-flows-and-guided-pathsSkill
- visual-emphasis-and-hierarchySkill
- wcag-accessibilitySkill
The file
We split this skill into the rules it gives an agent
7 rules. About 0 are common, 6 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈6
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 68
No lorem ipsum: Use real copy, or the closest draft of it.
- Distinctive: line 72
If you genuinely cannot use production data, generate the mock from the real schema and the real renderer, so its shape, formatting, and …
- Distinctive: line 78
When a visual stands in for what the product produces — a CLI run, a generated file, a report, a chart, an editor — reproduce the actual …
- Very specific: line 80
Use the product's real output structure: the same labels, the same ordering, the same tree/log/table shape, the same truncation rules ("+4 …
- Distinctive: line 94
Labels match content: Do not relabel one thing as another (a screenshot of system A captioned as system B.
- Distinctive: line 120
Keep claims conservative — under-promise in the visual, let the product over-deliver.
- Distinctive: line 129
When a design export and its own screenshot disagree, the screenshot wins.
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Authentic Product Representation
The fastest way to lose a visitor's trust is a product visual that is obviously staged. People have seen thousands of landing pages; they recognise a fabricated dashboard, a screenshot full of lorem ipsum, or a "terminal" that prints things the real tool never would. The visual that converts is the one that looks like it was screenshotted straight out of the working product — because it was, or because it was built to be indistinguishable from it.
The principle is simple: the visual is the product, not a poster of it. Every shortcut away from that — invented data, decorative chrome, a capability shown that does not exist — is a small withdrawal from the trust account.
This skill governs whether a visual is credible; what it has to accomplish in the page's argument comes from the product narrative framework in [[layout-paradigms-and-consistency]]. A fabricated proof point proves nothing.
Design With Real Content
Idealised placeholder content hides the problems your design will actually face.
- No lorem ipsum. Use real copy, or the closest draft of it. Real words have real lengths, real line-wraps, real awkward edge cases. Latin filler is uniformly tidy in a way your content never will be.
- Real names, real numbers, real lengths. "John Smith / $1,234" is a fantasy dataset. Use the longest realistic name, the account with 0 items, the title that wraps to three lines. If the design only looks good with perfect data, it is not finished.
- Design the empty, loading, error, and overflow states too — they are the product as much as the happy path. A screenshot that only ever shows the full, perfect state is selling a product that does not exist.
If you genuinely cannot use production data, generate the mock from the real schema and the real renderer, so its shape, formatting, and constraints match what ships.
Mirror the Real Output
When a visual stands in for what the product produces — a CLI run, a generated file, a report, a chart, an editor — reproduce the actual format, not a prettier reinterpretation.
- Use the product's real output structure: the same labels, the same ordering, the same tree/log/table shape, the same truncation rules ("+4 more").
- Pull the values from a real run where possible. If the visual claims to process
yourdomain.com, show what processingyourdomain.comactually yields — not another site's values relabelled. - Reuse the real rendering component if one exists. A faithful embed beats a hand-built lookalike that drifts from reality the moment the product changes.
- Breadth is more convincing shown than claimed. Listing the real categories the tool already covers ("also: borders, shadows, motion, components") communicates scope truthfully; a "coming soon" teaser for things that already exist communicates the opposite.
A reinterpreted mock is a maintenance liability and a credibility risk: it looks like the product until someone compares it to the product.
Keep It Truthful
The visual makes claims. Every one must hold.
- Numbers reflect reality. Stats, counts, and metrics are real or clearly illustrative — never inflated fiction presented as fact.
- Labels match content. Do not relabel one thing as another (a screenshot of system A captioned as system B; a domain in the command that does not match the results shown).
- Show only real capabilities. A demo that performs a feature the product does not have is a promise you will break on first use.
- "Soon" means soon, and rarely earns its place. A pending/soon badge on something that already ships is a lie; a pending badge on a real roadmap item is usually just noise in a product shot. Prefer showing what works.
Refuse Fabricated Marketing Chrome
A recognisable visual dialect signals "generated to look impressive" rather than "screenshotted from a real product." It reads as AI-slop and erodes the authenticity you are trying to build.
Avoid:
- Invented dashboards, fake charts, and decorative "terminals" that print marketing copy
- Gradient-filled text, sparkle/star accents, aurora glows, and glassmorphism used as a substitute for substance
- Typewriter/rotator headlines and faux-AI chrome (scanning lines, "thinking" shimmers) with no real function
- Pill badges and ornaments that exist only to fill space, especially in places where they clash with the medium (e.g. a label chip dropped into monospace output)
The restraint test: if an element does not exist in the real product, it does not belong in a visual that represents the real product. Decoration that carries no information is the first thing to cut.
When You Must Stage
Sometimes a literal screenshot is impossible (pre-launch, sensitive data, a composite view). Stage honestly:
- Build from the real schema, the real component, and realistic values.
- Match the product's real states and formatting, including the unglamorous ones.
- Keep claims conservative — under-promise in the visual, let the product over-deliver.
- Label clearly illustrative figures as such rather than passing them off as measured.
Photos on a Tinted Stage
A product shot on a tinted tile needs a transparent cutout. An opaque shot with a baked-in white background shows as a white box on the tint. Where cutouts do not exist, multiply-blend the photo onto the tile so the white vanishes.
When a design export and its own screenshot disagree, the screenshot wins. The export is the tool's opinion of the file. The rendering is what shipped.
Review Checklist
- Is there any lorem ipsum or obviously idealised placeholder content left? (There should be none.)
- Does the content use realistic lengths — the long name, the empty list, the three-line title?
- Are the empty, loading, error, and overflow states represented, not just the perfect state?
- When the visual stands in for product output, does it match the real format, ordering, and truncation?
- Do the values correspond to the input shown (right domain, right source), not another case relabelled?
- Is the real rendering component reused where one exists, instead of a hand-built lookalike?
- Do product photos on tinted surfaces use transparent cutouts or a multiply blend?
- Are all numbers, labels, and capabilities truthful — nothing inflated, mislabelled, or not-yet-real?
- Is there any "coming soon" framing on things that already ship?
- Is there fabricated marketing chrome (fake dashboards, gradient text, sparkles, decorative terminals) that should be cut?
- Could a visitor screenshot the real product and get the same impression this visual gives? If not, why not?
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

