Skip to main content
A skill is a reusable creative recipe — a named block of instructions the agent follows end-to-end when you invoke it. Think of it as a prompt template with teeth: structured enough that the agent executes it reliably, flexible enough that you can still hand it new subjects and contexts. Good candidates for skills:
  • A “hero shot” recipe that always produces the same composition, lighting, and finish
  • A “story reel” that always builds a 4-scene narrative from a single product
  • A “platform pack” that turns one idea into Instagram square + Reel + LinkedIn variants
  • A “brand audit” that runs a consistent visual analysis across saved inspirations
If you’re copy-pasting the same multi-step prompt more than twice, it belongs in a skill.

​
Where skills live

Open Brand Intelligence from the nav and switch to the Skills tab. This is your library of skills — every skill you’ve created shows up here, shareable across every Brand Kit in the workspace. From here you can:
  • Create a new skill (click New Skill)
  • Edit a skill’s prose, attached references, default style, and model pins
  • Search through skills by name as the list grows
  • Delete skills you don’t use

​
Two ways to create a skill

​
Write it yourself in the editor

Click New Skill. The editor opens as a two-column modal:
  • Left column — the workflow. A prose editor where you describe what the skill should do. Write it like a brief you’d hand a designer: what to produce, how many deliverables, any constraints, any style anchors. Type @ to mention any brand asset, color, palette, or kit — the mention becomes a reference chip attached to the skill.
  • Right sidebar — the settings.
    • Attached — the reference chips you’ve @mentioned. Click + Add to attach more. These are the concrete assets the skill leans on (a specific product photo, a logo, a style reference).
    • Default style — a one-line tone anchor the agent carries into every invocation (e.g. “editorial, matte paper finish, warm key light”).
    • Extra directions — a short list of rules or constraints that always apply (e.g. “never center the product; always leave margin top”).
    • Advanced (collapsed) — pin specific image or video models this skill should run on, overriding your workspace default. Leave on Default to follow the workspace.
The skill’s header icon is the thumbnail of the first attached reference — so a hero-shot skill looks like the product it produces, not a generic icon.

​
Ask the agent to build one for you

In Studio chat, select an image on your canvas and ask the agent to turn it into a skill (“save this as a reusable recipe”). The agent drafts a skill card in the conversation — prose + attached references + default style — for you to review. When the draft looks right, click Save Skill. The card flips into a confirmed “Saved” state and tells you exactly how to invoke it (Use /your-skill-slug in chat). The skill is now in your library.

​
Invoking a skill

Two ways to call a skill in Studio chat:

​
Slash command

Type / in the chat composer. A picker appears showing all your skills, filtered as you keep typing. Pick one, or hit Enter on the highlighted result. 
/hero-shot for the new soap bar, clean studio look
The slash command attaches the skill to the message. The agent reads the skill’s prose, default style, extra directions, and attached references as the brief for this turn — your free-text addition is the subject.

​
@mention

Type @ followed by a skill name. Works the same way as @mentioning a brand kit — the skill’s instructions become part of the agent’s context for that turn. 
@story-reel build a 4-scene launch narrative for the soda line
Slash commands are the faster path; @mentions are useful when you want to combine a skill with a brand kit mention in the same message (@Acme Brand @story-reel ...).

​
Plan-then-execute

For skills that produce multiple deliverables (like a storyboard recipe or a platform pack), the agent proposes a plan first — usually as a creative brief on your canvas. You review the plan, adjust aspect ratios and references if needed, and click Generate. The skill’s prose controls the plan; your confirmation controls execution. This means skills slot neatly into the Creative Brief flow. A skill that builds a storyboard produces a brief with the sequential toggle already checked. A skill that builds a platform pack produces a brief with one deliverable per platform — each with its own aspect ratio.

​
How aspect ratio and quality get chosen

Skills hold the creative intent — the style, the references, the directions. They don’t hold a fixed aspect ratio or quality. That’s intentional: the same /hero-shot skill should be able to produce a 1:1 feed post AND a 16:9 banner without you needing two near-identical skills. Here’s how it resolves:
  • For single-output skills (one image, no plan) — the agent asks you once for aspect ratio and quality in a single clarifying question, then generates. If the skill’s prose already specifies a ratio (“produce a vertical 9:16”), the agent uses that and skips the question.
  • For multi-deliverable skills (platform pack, storyboard) — the agent builds a brief with one deliverable per output, each with its own aspect ratio and quality row. You adjust inline, click Generate, and the brief’s server-side enforcement guarantees each image ships at its row’s settings.
This is why you’ll occasionally see a short “What aspect ratio and quality?” question before a skill runs. It’s a feature: one skill, every format.

​
Combining skills with brand kits

Skills are workspace-wide — one library, shared across every brand kit. @mention a brand kit in the same message and the agent combines the two: the skill provides the recipe, the kit provides the visual identity and voice. 
@Acme Brand /hero-shot a new packaging drop
The agent executes the hero-shot recipe using Acme Brand’s colors, fonts, voice, and assets as the creative constraints.

​
Editing skills

Skills are editable any time. Changes apply to future invocations immediately — already-generated work doesn’t change. If a skill is producing inconsistent output, tighten its prose: be more specific about deliverable count, style anchors, and required references. The editor opens with every attached reference already tokenized inline in the prose as mention chips, so you can see which assets the skill relies on at a glance.

​
Tips

  • Start from a prompt you’ve already perfected. If you have a multi-step prompt that keeps working, paste it into a skill’s workflow and give it a name.
  • Make skill names short. You’ll type them after a slash. /hero-shot beats /product-hero-shot-three-angles.
  • Keep each skill single-purpose. One recipe per skill. If a skill is drifting into “and also does X,” split it.
  • Write the skill prose in the agent’s voice. You’re writing the brief you’d give the agent. Be specific about what to produce; vague about exactly how, so the agent has room to use your brand kit.
  • Attach the concrete references. If the skill mentions “the character” or “the product,” make sure that asset is actually in the Attached panel — otherwise the agent has prose pointing at nothing.
  • Use Default style for the through-line, Extra directions for the rules. Default style = tone (“warm editorial, matte finish”). Extra directions = constraints (“never crop the logo, always leave margin top”).
  • Test each skill with two different subjects before relying on it. If the output differs wildly in quality between subjects, the prose needs tightening.