A skill that makes Claude build a moodboard the way a senior art director does, one stage at a time, with you deciding at each fork. It starts by pinning down the one decision the board must settle and who signs it off. It writes the feeling as a single sentence and three to five words, each paired with the word it must not become. Then it gives each of 9 to 16 frames one job and describes it as a photographer would, before finding it, searching for it or planning a phone shot. It pulls a measured colour strip and a texture strip from the frames, lays the board out for Figma, Milanote or Canva, and credits every image you did not make.
Inside are the working rules of a board that decides something. Fourteen feeling words are translated into light direction, colour temperature, distance and material. Frame jobs come with counts for 9, 12 and 16 frames, and seven tests run on the plan and again on the built board, including the squint test and a competitor's name in the header. Search strings are built from nouns, light and materials, never from feeling words. Open image licences, tool limits and image model settings sit in reference files with their sources and the date they were checked, and a worked example follows a fictional workshop brand through all seven stages.
It cannot take photographs or make images. For the frames nobody can find or shoot, it writes prompts for Midjourney, ChatGPT Images, Nano Banana or Adobe Firefly, sized to each cell on the board. It will not invent a creator, a licence or a Pantone number, and an image found through search stays marked unseen until you open it. Its notes on licences are not legal advice. It works best with code execution on, so it can measure colour from your images and hand you a starter file, and with web search on if you want it to look for frames. Ten minutes with your phone often gives it the best frame on the board.
Claude skill · version 1.0
Moodboard builder
Checking your access.
In the download · 7 files
- SKILL.mdThe method Claude follows in seven stages, from the brief and the feeling to frames, strips, layout, prompts and the final check, with an output template and one closing question per stage.
- README.mdHow to install it in claude.ai or Claude Code, how to start, and the licence.
- references/feeling-to-frames.mdThe feeling sentence, word pairs, fourteen feeling words translated into light, colour, composition, texture and people, frame jobs with counts for 9, 12 and 16 frames, edit mode and the board tests.
- references/sources-and-credits.mdWhere to look for each frame job, the terms of museum open access, Unsplash, Pexels, Wikimedia Commons and the Library of Congress, tracing found images, credit formats and rights labels, with sources and dates.
- references/strips-and-layout.mdColour and texture strips, tested code that measures colour and checks contrast, the grid for 9, 12 and 16 frames, and Figma, Milanote and Canva build steps and limits, with sources.
- references/image-prompts.mdWhen a frame earns a prompt, which tool suits which frame, the prompt paragraph, ratios per cell, Midjourney, ChatGPT Images, Nano Banana and Firefly settings, privacy defaults and how to check results.
Install it and begin
01Install it in Claude CodeTerminal mkdir -p ~/.claude/skillsunzip ~/Downloads/moodboard-builder.zip -d ~/.claude/skills/# Start it with /moodboard-builder. If this is your first skill, type /reload-skills once.On claude.ai there is nothing to unzip: upload the zip as it is. Every step is on Install and use a Claude skill.
02Your first message to ClaudeText Use the moodboard builder skill. The project is [PROJECT AND BRAND] and the work ends up in [SHOOT, IDENTITY, PACKAGING OR FILM] for [CHANNELS OR FORMATS]. It should feel like [A SITUATION, NOT ADJECTIVES]. I will build it in [FIGMA, MILANOTE OR CANVA]. My brief and references are attached. Start with what this board has to decide.Attach your brief, brand guidelines with colour codes and any images you already have, even a messy Pinterest export. Turn on code execution so it can measure colour and make a starter file, and web search if you want it to look for frames.
Worked example
Eight steps to a board that decides
- 01Write the one decision the board settles as a sentence that starts "After this board, we will know".A board with no decision attached grows to forty images, and the client picks the frame nobody meant.
- 02Turn the feeling into one situation sentence and three to five word pairs, such as "warm, not orange".The second word names the cliché the first word slides into, which is where boards drift.
- 03Give out the frame jobs before looking for anything: one anchor, one counterpoint, and light, texture, place, people and object frames between them.A frame with a stated job can be judged against it. A frame with no job can only be liked or disliked.
- 04Search with nouns, light and materials, never with the feeling words, and shoot what you can on a phone in ten minutes."Backlit linen curtain morning window" finds a frame, while "dreamy calm" finds the first page of stock. Phone frames are real and already yours.
- 05Pull the colour strip from every frame, weighted by the space each one covers, with the brand colours fixed first.A strip taken from the anchor alone matches one picture and misses the rest of the board.
- 06Place the anchor top left and the counterpoint in the far corner, and keep frames with the same job apart.Placement tells the viewer what to read first and shows that the odd frame out is a choice.
- 07Write prompts only for frames nobody can find or shoot, and tag each one G with the tool, model and date.A generated frame holds a quality for the board. It never stands in for the product, a real place or a real person.
- 08Run the squint, swap and stock tests on the built board, then ask the client whether this is the light for the shoot.A question about the work gets a decision you can act on, where asking if they like it invites taste.
What to do it with
- ClaudeFree tier, paid above it
A language model that writes, edits, reads long documents and runs tools, including agents that work inside a codebase.
Bad at It is agreeable by default, so it will improve a bad idea instead of telling you it is bad. Ask it to argue against you and the quality of the answer changes.
- Claude CodeSubscription
An agent that reads a whole repository, edits files, runs the tests and explains what it changed.
Bad at It is fast enough to make a large mistake quickly. Work on a branch, read the diff, and never let it push. Watch the starting mode: on Pro, Max and Team plans it opens in auto mode, which Anthropic's own list says may read a .env file and send the key in it to the API that key belongs to. Switch it off.
- FigmaFree tier, paid above it
The shared file where interface design happens: components, variables, prototypes and the handoff to code.
Bad at Its generative features make plausible screens rather than considered ones, and a plausible screen is the hardest kind to argue with in a review.
- MilanoteFree tier, paid above it
A board where images, notes and links sit together, built for the messy early stage.
Bad at Boards get big and stop being readable. It has no opinion about editing.
- CanvaFree tier, paid above it
Templates, brand kits and generation in one place, built so a non-designer can produce something acceptable.
Bad at Acceptable is the ceiling as well as the floor. Its templates are so widely used that a trained eye dates a deck in a second.
- MidjourneySubscription
Generates images from a written description, with the strongest default aesthetic of anything in the category.
Bad at That default aesthetic is the problem as often as it is the draw. It has a house look, it pulls everything towards it, and getting a plain honest photograph of a real product out of it is harder than it should be. Text inside an image is unreliable.
- GPT ImagePay per use
Generates and edits images with unusually good instruction-following, and can hold text inside a picture better than most.
Bad at Faces drift between generations unless you feed a reference every time. Skin can come out plasticky at default settings, and a heavy fidelity setting costs many times more and takes many times longer.
- Adobe FireflyFree tier, paid above it
Adobe's generative models, trained on licensed and public-domain material and wired into Photoshop and Illustrator.
Bad at Raw image quality trails the leaders. What you are buying is the indemnity and the fact that it sits inside files you already have.
- GeminiFree tier, paid above it
Google's assistant, with a very long context window and tight links to their other products.
Bad at Behaviour changes between the app, the API and the versions inside other Google products.
Why it matters
Asked for a moodboard, a chat returns a row of adjectives and a pile of stock-like pictures, and the client fixes on the one frame nobody meant. A board works when every frame is there for one stated reason, the palette comes from the frames themselves, and every found image can be traced to its maker. This skill makes Claude settle the decision first, choose each frame against agreed words, and carry a credit for every image from the moment it enters the plan.
- Type
- technique
- Period
- ongoing