Using the TrendAI™ Image Library

Everything in this library is approved for use — search or browse, then take what you need: download a file, paste a CDN URL that keeps working, copy the artwork itself, or let an AI assistant pull assets straight into a conversation over MCP.

Using the library

Find an asset and use it correctly — in a deck, on a web page, or in an email.

The three libraries

The sidebar has one section per kind of asset. They look similar but hold quite different things, and what you can do with each differs.

Section	What's in it	Comes in
Icons	The TrendAI icon set — product, threat and UI concepts, drawn as line art.	SVG and PNG, in three colour treatments, some with an alert or shield overlay.
Images	Licensed stock photography, categorised (Technology, Business, Places, Industry, Lifestyle, Healthcare, Sports).	Four rendered JPEG sizes. The licensed original is download-only.
Logos	The TrendAI logo and brand lockups — horizontal, vertical, stacked, tagline, Vision One.	SVG/PNG for screen, plus AI/EPS/PDF print masters in CMYK, Pantone and RGB.

Finding an asset

Type in the search box at the top. Search is deliberately forgiving and runs in three passes, best matches first:

Exact — your words appear in the name or tags. Predictable, always first.
Meaning — nothing matched the words, so it looks for assets that mean the same thing. "phishing email" finds Business Email Compromise; "party" finds the celebration photo.
Typos — still nothing, so it tries near-misses. "firewal" finds Unsecured Firewall, "amail" finds email.

Wrap a phrase in "double quotes" to switch all of that off and match literally.

The chips above the grid are categories, each with a count scoped to whatever you've typed — so the number tells you what you'd get if you clicked, not how big the category is overall. Scroll the grid to load more; the slider at the bottom right changes how large the tiles are.

Opening an asset

Click any tile. A panel opens on the right with a preview and, depending on the kind of asset, tabs for choosing which version you want.

	First row of tabs	Second row
Icons	Variant — Color, Reverse-Red-White, Reverse-White (the "Reverse" ones are for dark backgrounds).	Mark — an overlay on the same icon: base (none), alert, or shield. Only shown when the icon has more than one.
Logos	Treatment — Full-Color, Black, White, Inverted.	Print masters are listed separately, per colour space (CMYK, Pantone, RGB).
Images	Size — thumbnail, small, medium, large.	—

Everything below the preview follows the tabs you've picked, so switching a tab updates every URL and snippet at once. The − 100% + control zooms the preview only, to inspect fine detail; it doesn't change what you download.

Three ways to use it

The panel offers the same asset three ways. They're not interchangeable — picking the wrong one is how a logo ends up as a broken image in someone else's deck.

If the asset needs to…	Use
travel inside a file you send someone	Download
load over the network when a page opens	Embed
work with no network at all	Copy the content

Download a file

For a deck, a document, or print.

Gives you the file. Use this whenever the asset has to travel inside the document: PowerPoint, Word, InDesign, a PDF going to a printer.

Images come in four sizes, measured on the longest edge, and are never scaled up past the original:

Size	Longest edge	Good for
thumbnail	300 px	a list, a contact sheet
small	800 px	inline in a document, email
medium	1600 px	a slide, a web hero
large	2400 px	full-bleed, print

Images also offer web or print. The pixels are identical — the only difference is the DPI recorded inside the file (72 vs 300), which is what tells InDesign how big to place it. Pick print if it's going into a print layout, web for anything else.

The licensed original of a stock image is never on the public CDN. It's available to download only to people with access to restricted items, because that's the file the licence actually covers.

Embed with a URL

For anything that loads over the network.

The CDN URL rows and the ready-made <img> tag. Use these on a web page, in an email template, a dashboard, a wiki — anywhere the asset can be fetched when the page opens. Nothing to attach and it updates if the asset is ever re-cut.

These URLs are cached hard and can carry a ?v= version stamp, so a page you ship today keeps rendering exactly what you tested.

Copy the content

For offline documents and email.

SVG markup pastes the vector artwork itself as code — sharp at any size, recolourable in CSS, no network request. Offered for SVG assets only, since a photo has no markup.

base64 pastes the image inlined into the tag itself. It's bulky, but completely self-contained — the right choice for an email that must render with images blocked, an offline document, or a tool that can't reach the CDN. The row names the format it gave you (PNG, JPG…).

If the embed rows are missing and a short note sits in their place, the asset is restricted — see below. Download and copy-content still work; a public URL deliberately doesn't exist, so pasting one would only break later, for whoever opened the page.

The purple padlock

Nearly everything you can see here is simply ready to use, with nothing on the tile to think about. The exception is a purple padlock:

Restricted — this asset isn't public, and you're seeing it because you've been granted access to it. You can view it, download it and copy its content; what you can't do is embed it, because it deliberately has no public URL. Purple means restricted everywhere in the app, including the padlock filter in the toolbar, which narrows the grid to just these.

If you haven't been granted access to anything restricted you'll never come across this — restricted assets aren't listed for people who can't use them.

AI assistants & API

Connect this library to Claude (or any MCP client) and let it fetch brand assets for you.

What MCP is

MCP is a standard way to give an AI assistant real tools. Connect this library once and your assistant can find and use brand assets itself — "put the right Trend logo in this deck", "find an icon for zero trust" — instead of you searching, downloading and pasting.

Connecting an assistant

Open About → Connect to Claude. Two options are offered there:

As a connector (recommended) — paste one URL into Settings → Connectors → Add custom connector in Claude.ai or Claude Desktop and sign in with your Trend account. Nothing to copy, nothing to rotate.
As a config entry — for Claude Desktop or any other MCP client, a JSON snippet to paste into its config, with a token you choose the lifetime of.

The tools it gets

Two tools, deliberately — a small surface an assistant can use well, rather than the whole REST API:

Tool	What it does
`search`	Finds assets by keyword, with the same three-pass ranking the search box uses. Can be narrowed by type (icon, logo, photo), category, tag, and for icons and logos the variant and format it should point at. Each hit comes back as `{type, slug, path}` plus one `host` for the whole response — join them and you have a working, embeddable URL, so simple requests need no second call.
`get`	Full detail for one asset when the first answer isn't enough — every variant, mark and format, its categories and tags, and the print masters it has (as names, not links).

For photos, the assistant can swap the size itself: the response lists the available sizes, and changing the last part of the path from /medium.jpg to /large.jpg is all it takes.

What to expect (and not)

The MCP surface only ever returns published, unrestricted assets and only the formats a browser can render. Drafts, deprecated, retired and restricted items are invisible to it, as are print masters (.ai, .eps) and licensed originals — and a hidden asset answers exactly like one that doesn't exist, so nothing is revealed by asking. Anything an assistant hands you from this library is something you were already free to publish.

REST API

Everything the app does is a documented HTTP API, and the MCP tools above are generated from that same schema, so the two can't drift. Browse it in Swagger UI or ReDoc.

Editing assets Marketing editors

Add, change, publish, restrict and remove assets — and what each of those does downstream.

How editing works

You can create assets, upload files, change what's published and control who can reach what. This half of the guide explains what each of those actually does — several of the controls have consequences that aren't visible from the form.

Two habits worth having: metadata is staged, files are not. Editing a name, description, categories or tags does nothing until you press Save changes, and Cancel throws it away. Uploading, replacing or deleting a file happens the moment you click it — Cancel won't undo a deletion.

What an item is

An item is one asset — one icon, one photo, one logo lockup — and it owns a set of files. "TrendAI Logo — Horizontal" is a single item holding 35 files: four treatments × SVG and PNG for screen, plus EPS and PDF print masters in three colour spaces, plus three Illustrator artboards.

That's the unit everything else attaches to:

Name — what people read. Editable at any time.
Slug — the URL-safe version of the name, derived automatically. Rename the item and the slug follows, but the old one keeps working forever, so nothing anyone has already pasted breaks.
Id — a 26-character code, generated once, never changes. It's what the app's own URLs use, which is why two assets can safely share a name.
Files — each with its own variant, mark, format and access. Filenames are built for you from those (slug_variant_mark.ext), never typed, so they stay predictable.

Item types and their axes

Every file is filed under up to two labels. Variant is the visual version. Mark is a second distinction within a variant, and it means something different per type — which is the one thing about this model that catches people out.

Type	Variant	Mark	Files per item
Icons	Colour treatment: Color, Reverse-Red-White, Reverse-White	An overlay: base (none), alert, shield	Up to ~30 (treatment × overlay × SVG/PNG)
Logos	Treatment: Full-Color, Black, White, Inverted	A colour space, on print masters: cmyk, pantone, rgb	~35
Images	None	None	Exactly one — the licensed master. Every size shown in the app is rendered from it on demand.

The lists are fixed, not free text — you pick from a dropdown when uploading. That's what keeps filenames derivable and the variant tabs consistent across hundreds of items.

Adding a new asset

Click + New icon / image / logo in the toolbar, type a name, create. The item is a Draft and the grid switches to Draft so you can see it.
The editor opens on the empty item. Add files from Add a file — drop them on the box or choose them.
For each upload, pick its variant and mark from the dropdowns, and its access.
Fill in categories and tags — an item nobody can find is not much better than one that doesn't exist.
Set the status to Published when it's ready, and Save changes.

Access defaults to Restricted on every upload, on purpose. Publishing is meant to be a deliberate act. The cost of getting that wrong in the safe direction is a file nobody can reach yet, which you notice and fix in a second; the other direction puts an unreleased asset on a public CDN, and a URL that has been fetched can't be un-fetched.

Only formats a browser can draw are worth marking Public — SVG, PNG, JPG, WebP, GIF, AVIF. AI, EPS and PDF are print and source formats: keep them Restricted, which is where the download route serves them from anyway.

A note on icons specifically: SVG is the format to aim for, PNG as a fallback. A JPG can't have a transparent background, so an icon uploaded as a JPG will show a solid rectangle wherever it's placed.

Uploading and replacing files

In edit mode, every file the item owns is listed in two groups — Public (served from the CDN) and Restricted (sign-in required) — with its variant, mark, size and an access switch.

The group headers count what's in them, so you can see at a glance whether an item's split is what you expect.
A warning on the public group means something in there isn't a web format — an EPS or Illustrator file marked Public. It's flagged because that's a print master about to be published, which is nearly always a mistake.
Replace swaps the bytes and keeps the filename, so every URL already in circulation now serves the new artwork. That's the right tool for a corrected file — and the wrong one if you meant to publish a different asset.
Delete removes that one file immediately. There's no undo and Cancel doesn't cover it.
Renaming an item renames its files, and the old filenames keep redirecting to the new ones, permanently. Links don't rot.

Naming, categories and tags

Name: What people read, and what the slug is derived from. Renaming is safe: the previous slug keeps resolving for good.
Description: Optional context — usage notes, what makes this variant the right one. Shown in the panel.
Categories: The chips people browse by. A controlled vocabulary in Title Case ("Files & Data", "Cloud & Infra"), so type them as they appear elsewhere — a near-miss creates a second, nearly identical chip. Keep names short: the chip row wraps, and long names push everything else down.
Tags: Free text, lowercased, one word or short phrase per chip. This is the single most useful field you can fill in: tags feed both exact and meaning-based search, and they're browsable as facets. Add the words people will actually type, including synonyms the name doesn't contain.

There used to be a separate "Keywords" box. It did exactly what Tags does, with nothing on screen to distinguish them, so the two were merged — every old keyword is now a tag. There's one free-text field, and it's Tags.

Changing status

Status answers "how finished is this, and who should be using it". It's set from the row of buttons in the editor, and filtered from the matching row in the toolbar. Each one behaves differently on the public CDN, which is the part worth knowing before you switch anything.

Worth knowing before you read on: these badges are only ever visible to you and other editors. For everyone else the library contains published assets and nothing else — draft, deprecated and retired items are dropped from their search results, their category counts and their browsing entirely. Changing a status doesn't move an asset between labels in someone's view; it moves it in or out of their view altogether.

Draft Draft

Who sees it: In this app, editors and admins only. It's absent from everyone else's search results, from the public API listings, and from MCP.
Public CDN: Nothing is discoverable — but the CDN checks the file's own access, not the item's status, so a file you explicitly mark Public is still fetchable by anyone who has its exact URL. Uploads default to Restricted for precisely this reason: leave them Restricted until you publish, and a draft really is unreachable.
Use it when: You're still assembling the item: it's the status a new item starts in, and where it should stay until the files are right.

Published

Who sees it: Everyone, and AI assistants over MCP.
Public CDN: Serves normally, cached hard.
Use it when: The asset is approved for general use. This is the only status that appears in a normal person's search results.

Deprecated Deprecated

The point of it: Stop offering it, without breaking anything that already uses it. Those are two different questions, and Deprecated is the status that answers them differently: nobody browsing the library is shown the asset any more, so it can't be picked up for anything new — while every page, deck, email template and service that already embedded it keeps working exactly as before, at full quality, indefinitely.
Who sees it: Editors and admins, badged. It's gone from everyone else's search and browse results, and from MCP, so it can't be chosen again.
Public CDN: Still serves the real artwork, at full quality — no placeholder, no redirect, no expiry. Existing embeds are untouched.
Use it when: Something newer replaces it but the old one is legitimately still out there. This is the gentle option, and usually the right first move: retire it later, once you know the remaining uses are gone.

Retired Retired

Who sees it: Editors and admins, badged. Invisible to normal search and to MCP.
Public CDN: The URL still answers, but with a generated placeholder image instead of the artwork. Anyone still embedding it sees a placeholder appear where the asset was.
Use it when: The asset must stop being used — off-brand, expired licence, wrong product name. The visible placeholder is the point: it surfaces the remaining uses so they can be fixed, instead of leaving broken images or, worse, an off-brand logo still quietly serving.

Retired is the reversible way to remove something. The item, its metadata and its files all stay; switch it back to Published and everything resumes. Deleting is permanent and takes the files with it. If your instinct is "get rid of this", reach for Retired.

Restricting an asset

Status is about how finished something is. Restriction is about who's allowed to have it — an unreleased product logo, a photo whose licence doesn't cover public use, print masters that were never meant to leave the brand team.

There are two separate switches, at two different levels.

The item switch

A Restricted toggle in the editor, which reveals a list of people you can grant access to. When it's on, the whole item is off the public internet.

The per-file switch

Every row in the files table is Public or Restricted. This is how one item can publish its SVG and PNG for anyone while keeping its Illustrator masters internal — which is exactly how every logo is set up.

The item switch wins. Turn on Restricted at the item level and every file goes off the CDN, including files still marked Public in the table. Their access isn't changed — it's overridden while the item flag is on, and comes back when you turn it off. So an item can look like it has public files and serve none, and that's working as intended.

Who can see a restricted asset

Any one of these is enough, and they add up:

Platform admins — always.
Marketing editors — you curate the library, so you can see all of it.
The "Access to restricted items" list — a platform-wide list an admin manages in Settings. Being on it is not editing permission; it's the people entitled to restricted assets and licensed masters.
Per-item grants — the email list on that one item. Use this for "one person needs this one asset": it grants nothing anywhere else.

Open a restricted item and the panel shows a purple callout naming how many people can reach it; expand it for the computed list — the platform list and that item's own grants combined, so you can see the real answer rather than the two halves.

What each person experiences

The same restricted asset, side by side, from both sides of the permission:

	Someone with access	Everyone else
In the grid	The item appears, with a purple padlock over the real artwork.	The item isn't there at all.
The preview	The actual artwork, fetched through a signed-in route.	—
Download & copy content	Both work.	—
CDN URL / <img> tag	Gone, with a note explaining why.	—
A direct link to it	Opens.	"Not found" — the same answer as an asset that doesn't exist, so no one learns it's there.

The embed rows disappear on purpose. A public URL for a restricted asset can't be made to work, so offering one would only hand someone a link that fails later, on whoever opens the page rather than on the person who copied it.

Restriction is enforced by the server on every request, not by hiding buttons. A restricted file has no working public address at all, whatever anyone types.

Deleting an asset

At the bottom of the edit form is a Danger zone. Deleting removes, permanently and immediately:

every file the item owns, in every variant and format;
the item's metadata;
its entry in the search index and in meaning-based search;
any cached renditions made from it.

Editors can only delete an item that has no files — the case this exists for, a mis-created empty draft. An item with files can only be deleted by a platform admin, and the button says so rather than failing when you press it.

There is no undo, and no copy kept. For anything that has ever been published, use Retired instead — it stops the asset being used, keeps the files, and can be reversed.

Administration Platform admins

Roles and access lists, the variant vocabulary, and the environment watermark.

Who can do what

Settings (the gear beside the toolbar filters) holds four email lists. Three of them are roles; the fourth is deliberately not.

List	Grants
Seed admins	Everything. Comes from deployment configuration and is read-only here — which is what makes it impossible to lock every admin out of the app from inside the app.
Additional admins	The same as seed admins: these settings, the watermark controls, deleting items that still have files, and this list itself.
Marketing editors	The whole editor guide above — create, upload, publish, restrict, delete empty items. No platform settings.
Access to restricted items	Not a role. It's who may see and download restricted items and licensed masters (.ai, .eps, photo originals). Nothing about editing. Admins have it inherently; an editor can browse restricted items without it. Put people here when their job needs the masters — a designer, an agency contact — not when they need to curate.

For one person who needs one asset, don't add them here — grant it on that item, in its own access list. Narrower, and it expires with the item rather than outliving the reason.

Changes take up to about 30 seconds to reach every server, and someone already signed in may need to sign in again to pick up a new role. A person reporting "you added me but I still can't see it" a few seconds after you saved is expected, not a fault.

Variants and marks

The variant and mark dropdowns editors pick from are a controlled list per asset type, managed in Settings. They're a fixed vocabulary rather than free text for one reason: filenames are derived from them (slug_variant_mark.ext), so a typo in a free-text box would produce a permanent, wrongly-named file and an extra tab in every affected item.

Adding a value makes it available on the next upload. Removing one does not rewrite existing files — they keep their variant and keep working; the value simply stops being offered for new uploads. So removing a value in active use leaves items whose tabs mention something the dropdown no longer has, which is confusing rather than broken. Prefer adding.

Environment watermark

Non-production environments stamp a diagonal label across served icons and images, so an asset pulled from staging can't be mistaken for a real one and quietly shipped.

It can be switched off for an hour at a time when you need to check artwork properly, and it restores itself — there's no way to leave it off by forgetting, which is the whole point of a timed suppression rather than a toggle.

The three libraries

Finding an asset

Opening an asset

Three ways to use it

Download a file

Embed with a URL

Copy the content

The purple padlock

What MCP is

Connecting an assistant

The tools it gets

What to expect (and not)

REST API

How editing works

What an item is

Item types and their axes

Adding a new asset

Uploading and replacing files

Naming, categories and tags

Changing status

Draft Draft

Published

Deprecated Deprecated

Retired Retired

Restricting an asset

The item switch

The per-file switch

Who can see a restricted asset

What each person experiences

Deleting an asset

Who can do what

Variants and marks

Environment watermark

TrendAI™ Image Library and CDN

Developer resources

REST API

MCP

Project & contact

Connect to Claude Desktop

Settings

Environment watermark

Access

External access