AgentStack
Browse Sign in
Browse Why AgentStack Sell Docs
Sign in
SKILL verified MIT Self-run

Collateral Design

skill-zstvns-design-skills-collateral-design · by zstvns

Apply a brand's defined visual system to any fixed-format piece that isn't a website — decks and presentations, one-pagers and sales sheets, social posts and graphics, blog/OG thumbnails, business cards, brochures, flyers, posters, banners, billboards, and paid ads. Use when the user mentions 'design a deck,' 'pitch deck,' 'sales deck,' 'presentation,' 'slides,' 'one-pager,' 'sales sheet,' 'leave…

No reviews yet
0 installs
0 views
view→install

Install

$ agentstack add skill-zstvns-design-skills-collateral-design

✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

View the full security report →

Verified badge

Passed review? Show it. Paste this badge into your README, it links to the public security report.

AgentStack Verified badge Links to your public security report.
[![AgentStack Verified](https://agentstack.voostack.com/badges/verified.svg)](https://agentstack.voostack.com/security/report/skill-zstvns-design-skills-collateral-design)

Reliability & compatibility

Security review passed
0 installs to date
no reviews yet
yesterday

Declared compatibility

Claude CodeClaude Desktop

Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.

Preview Execution monitoring

We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.

How agent discovery & health will work →
Are you the author of Collateral Design? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

Collateral Design

You apply a brand's defined visual system to the collateral family: decks, one-pagers, print pieces, and ads. This is an Application skill — the color roles, type rules, motif, and mark from creative-direction and logo-design become finished, fixed-format artifacts. It sits beside web-design: same identity, different mediums.

Read design-principles first. Everything here is that skill meeting a page that can't scroll, can't reflow, and often can't be revised after it ships. Hierarchy is the job; restraint protects the message; coherence is not repetition — a deck, a sell sheet, and a feed ad can each look different and still be unmistakably one brand.

> Position in the sequence: after creative-direction (the system), logo-design (the mark), and alongside web-design — an Application skill. It consumes the identity; it does not define one. The medium is not one surface but a family of fixed-format deliverables, each with hard constraints web doesn't have: a set trim size, a projector's contrast floor, a thumbnail's legibility, a fold that reorders reading — and no second chance. The piece lands in one shot or not at all.

The Thing That Makes Collateral Different

Web reflows, responds, and ships in commits you can revise tomorrow. Collateral does none of that. Three constraints run through every format here and shape every decision:

  • Fixed format. A 16:9 slide, a US-Letter sheet, a 1080×1350 ad, a tri-fold's panel — the frame is set before you start, and the content is designed to it, not poured into whatever height it needs.
  • One shot. No hover to reveal, no scroll to continue, no "fold" to push detail below. A projected slide has ~3 seconds; a feed ad has ~1. What doesn't land immediately doesn't land.
  • Exported, not live. Collateral ships as a static artifact — a PDF, a sized image, a deck — not a surface you revise tomorrow. It has to stand alone: a deck read with no presenter, a one-pager with no scroll, and — when it's print — CMYK on stock with bleed, not RGB in your editor. Most collateral is screen-first; print is one destination, not the default. If it only works in your tool, with you narrating, it isn't finished.

Name the format and its constraints before you design — that's this skill's version of the front door.

Front-Door Diagnosis — Settled Brand, or Rework?

Before anything, diagnose where the brand is. Open .agents/design.md, .agents/brand.md, and .agents/product-marketing.md — by this stage all three should exist. Then ask one question:

> Are we applying a settled brand, or reworking one from scratch?

  • Refine / apply the existing system ("we have our brand, we need a deck") → Optimization or light Evolution. Proceed — this is the skill's home.
  • The system is being defined or changed (full rebrand, heavy Evolution, or the new type/color/graphics aren't recorded yet) → stop and route back to creative-direction. Do not design collateral on an undefined system — you'll produce one-off art that nothing else will match, which is the opposite of collateral's job.

**If .agents/design.md is missing, that is the "go back" signal. If .agents/product-marketing.md is missing or the piece has no copy, route to the marketing library's copywriting (or sales-enablement for a sales asset's substance) first: the piece is built around the message, never the reverse.** Routing back holds even if the user already has a logo — a mark without the system around it isn't enough to keep a deck, a sheet, and an ad coherent.

Copy First — This Is the Gate, Not a Preference

Weak collateral is usually a copy problem wearing a design problem's clothes. No amount of composition rescues generic lines; conversely, get genuinely good copy and the design gets good almost on its own, because there's finally something worth giving emphasis to. So this is a gate, not a nicety: if the copy isn't strong, stop and get it written — route to the marketing library's copywriting (or sales-enablement for a sales asset's substance). Placeholder-grade copy produces placeholder-grade design, and you will not be able to fix it downstream.

The exception: pieces with no real message to carry. A business card, a profile avatar, a letterhead, a folder — the content is a name, a role, a URL. These are pure identity applications and can be designed immediately, without waiting on copy. Everything that argues something — one-pagers, decks, social posts, ads, thumbnails — waits for the words.

Message First — Especially Here

Collateral exists to carry a specific message to a specific moment: a deck persuades a room, a one-pager survives being left on a desk, an ad interrupts a scroll. The single most important input is what it needs to say and to whom. Read product-marketing.md (audience, voice, positioning) and the actual copy. Your job is to give that message proper emphasis, joy, and delight — not to build a template and drop text into the boxes. Template-first is the fastest tell of assembled collateral: the deck where every slide is a title and three bullets, the sheet that's a wall of equal-weight paragraphs.

If the copy isn't written, get it written first. Size follows content, not a grid — one killer proof point deserves a whole slide; a feature list deserves consolidation, not a cramped two columns.

Apply the System — By Role, Across the Family

Pull the system from .agents/design.md; don't re-derive it, don't invent alongside it. The same role discipline as web-design, tuned for fixed formats:

  1. Color by role, in the right color space. Background, text (primary/secondary), action/accent, support — five stops, not ten. For anything going to print, work in CMYK and check the translation — brand RGB/hex often shifts on press (see [references/print.md](references/print.md)). For decks and screen-PDF, RGB is fine, but mind projector contrast: subtle low-contrast pairs that read on your monitor die in a bright room.
  2. Type by role, for contrast not competition. Display, heading, body — each with a job. Bold is not the default; body is a reading weight. Fixed formats punish under-scaled type: a deck read from the back of a room and an ad seen at thumbnail size both demand bigger, more decisive type than a website. Use the licensed brand faces; embed them in the file (PDF/PPTX) so they survive the handoff.
  3. Use the REAL assets from .agents/design.md — never a stand-in. This file (plus logo-design's output) holds the actual logo files, brand colors, licensed typefaces, and motif established upstream. Pull them; do not approximate them.
  • Never substitute an icon-font glyph or a lookalike shape for the logo. A Material Symbols / Lucide / emoji icon standing in for the mark is the single most damaging shortcut in this skill — it silently replaces the brand's most controlled asset with a generic one. If you can't locate the real file, stop and ask rather than approximating.
  • Check you have the current version. Identities get revised; a mark from an earlier round may be stale. Confirm the logo, palette, and type in design.md are the latest before applying them across a whole family — a full set built on a superseded mark is a full set to redo.
  • The mark has a canonical path — place the file, don't prepare one. logo-design delivers locked files to a fixed location and registers them in design.md's Logo section. Read the table there and place the file by name:

| Need | File | |---|---| | Default, everywhere | .agents/assets/logo/logo-lockup.svg | | Mark alone (tight space, avatar, favicon) | logo-icon.svg | | Dark ground | logo-lockup-inverse.svg | | One-colour (print, etch, single-ink) | logo-lockup-black.svg · -white.svg · -gray.svg | | Square / rounded | avatar.svg · app-icon.svg | | Surfaces that can't take SVG | raster/ |

A ground is a variant choice, not a colour edit. Light piece → the primary; dark piece → the inverse or white file. Do not make one asset inherit currentColor so the surrounding CSS decides its colour — that's not "one asset working everywhere," it's handing the brand's most controlled asset to whatever color happens to cascade in, which is how it ends up tinted to an accent. currentColor is legitimate only inside a file the identity designates as the one-colour variant, where the single ink is the point.

  • If the variant you need doesn't exist, route back to logo-design and have it added. Never improvise one at the point of use — not by recolouring, not by inverting in CSS, not by exporting your own.
  • **Read which mark was chosen, don't just grab a file. A logo exploration leaves behind many candidates — three families, a shortlist, several wordmark experiments. Find the stated recommendation / sign-off and use that one. Picking whichever file looks plausible (or whichever the presentation happened to reference in a layout) is how a whole collateral family gets built on a rejected concept. If the recommendation isn't recorded, or the assets are still placeholders, stop and ask** — this is not a guess worth making, because every piece in the family inherits the error.
  • The logo is ONE locked element. Use it exactly as delivered. Mark and wordmark come pre-locked in fixed proportion and fixed relationship — you place that single asset, you never rebuild the lockup out of parts, and you never re-space it. A logo must behave identically in every medium. Specifically, never:
  • recolor it — use only the delivered color variants (full-colour / inverse / white / black / mono). Tinting the mark to match an accent is a hard no.
  • box it — no container, chip, rounded square, or badge around the logo unless the identity itself defines one.
  • crop it — give it its full artwork bounds and clear space; a mark clipped by its own frame or viewBox is broken. Check the delivered viewBox actually contains the whole drawing.
  • re-declare its coordinate space — the delivered file's viewBox is part of the asset, not a layout choice. Never set a second one at the place of use: an ` wrapping a of a nests two viewports and silently re-maps the artwork, and stripping the symbol's own viewBox renders a fragment. Either way you get a subtly wrong mark on **every piece in the family**, and because it still reads as *a* logo it survives review after review. Size the placement with width/height (and flex-shrink: 0, since a viewBox-less ` has no intrinsic ratio for a flex parent to respect) — never with coordinates.
  • rebuild or re-proportion it — no setting the wordmark yourself next to the symbol, no adjusting the gap, no substituting a different cut per format.
  • Same for color and type: exact brand values and licensed faces from design.md, not visually-similar substitutes.
  • Diff your render against the delivered file — visually, side by side, large. Render the mark as you've placed it at ~200px next to the delivered artwork and confirm they are the same drawing: same stroke weight all the way round, same gap, same proportions. Then check it again at the smallest size you ship it. Verifying a number — the bounding box, the file hash, the path data — proves nothing about what renders, and a mark can be measurably "contained" and visibly distorted at the same time. If you never put your output next to the source of truth, you have not checked the logo.
  • If the mark's detail collapses at the sizes collateral actually needs, that's a finding — not something to fudge. A card, a slide footer, and an avatar are small; a mark with fine interior detail (thin ticks, knurled edges, hairlines) can turn to mush there. Report it back to logo-design and ask for a simplified small-size cut rather than shipping a muddy mark or quietly redrawing one yourself. Collateral is usually where this gets discovered, because it's the first place the mark is used small and repeatedly.
  1. The mark is present but rarely the hero — and use the right cut of it. A logo belongs on collateral (slide footer, sheet corner, ad sign-off) sized for recognition, not dominance; a cover or pure brand piece is the exception — and on those, size it like the hero it is. A footer-scale mark dropped into the middle of a closing slide or a cover is its own error: if the mark is the only thing on the surface, undersizing it makes the piece look unfinished rather than restrained. Give it clear space; never stretch, recolor off-system, or drop it onto a busy field. Inside a constrained shape — a circle, an avatar, a small badge — use the mark alone, not the full lockup. Nesting a mark-plus-wordmark inside a ring or a tiny container crowds both; pick the symbol and let it breathe. And don't add the logo where it earns nothing: on a piece already carrying the brand's motif and type, a redundant mark is clutter.
  2. The motif carries the brand across the family — with restraint where the message must win. The signature element (pattern, texture, recurring detail) is how a deck slide and an ad read as the same brand without being the same layout. This is where distinctiveness must survive application (see design-principles). Calm it wherever the message leads — a cover, a hero stat, an ad's one line — and let it work harder on connective, lower-stakes surfaces.
  3. One accent, reserved. Whatever signals "the point / the action" — the one CTA on an ad, the key number on a slide — stays scarce. Sprinkled on every heading and box, it stops meaning anything.

Compose the Frame — Dynamic, Not Filled

A fixed frame has no scroll to rescue a flat layout and no interaction to add interest, so composition carries most of the load. Full method in [references/composition.md](references/composition.md):

  • The canvas is a hard limit — content never exceeds it. Headlines, stats, quotes, cards, CTAs must sit entirely inside the safe area; only decoration (washes, patterns, oversized motifs, bleeding photos) runs off the edge. Clipped content is a failure, not a rough edge — and the fix is to shrink or cut the content, never to let it spill. Check all four edges and the bottom of the last element before calling a piece done.
  • The logo goes at the BOTTOM, not the top. The top of the frame belongs to the message. The mark signs off at the bottom (deck covers and pure brand pieces excepted), and social posts get no logo at all — the handle and avatar already brand them. On an identity piece this decides the whole layout: on a business card the person's name is the message, so the name leads at the top and the logo signs off at the bottom — not a wordmark banked at the top with the name pushed under it and a dead band of nothing in between.
  • Alignment is exact, spacing is systematic, punctuation hangs. Off-centre is sloppiness; a pull-quote's opening " must hang outside the measure so the text's left edge stays true. Run the craft pass with frontend-design / web-design-guidelines — those principles apply verbatim to a fixed canvas. Overlap only where it creates depth or connects two things.
  • Don't port web components into print. A browser frame, nav bar, app card, or form field pasted onto a sheet reads as a screenshot and usually drags the wrong typeface in with it. Share the system with the site, not its components — and set everything in the brand's actual face.
  • A bad image is worse than no image. Generic stock cheapens the piece; if you don't have imagery that's specific and good, solve it typographically or graphically instead.
  • Light first, dark second. Default to a light composition, especially for print — a near-black printed piece is rare, drinks ink, scuffs, and goe

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet, be the first.

Versions

  • v0.1.0 Imported from the upstream source.