How to use the OpenClaw Brand Identity agent skill

What is an agent skill, and what does this one do?

An agent skill is a folder containing a SKILL.md file and optional reference files. When installed, an agent can read these instructions and follow them for matching tasks. This skill is named openclaw-brand and is published in the openclaw/clawhub repository under .agents/skills/openclaw-brand, under the MIT licence.

When a task changes the brand itself, such as a logo variant, a presentation, a social graphic or launch material, the agent follows OpenClaw's brand rules. It uses the defined palette (ink and warm-paper neutrals, coral as the primary brand colour, sea glass as a restrained secondary accent), sentence-case headings, and Switzer-compatible sans-serif type for interface and body copy. It is told not to rotate, distort, outline or add effects to the logo, and not to use generic gradients, decorative blobs or ornamental glow in place of content.

• Reads identity.md for palette, typography, logo and voice before working on brand artefacts
• Reads asset-rights.md before copying or redistributing any brand file
• Uses semantic design tokens when the output is code
• Reports the logo clearspace check as unavailable unless approved guidance defines it, rather than guessing a measurement

Install the openclaw-brand skill

There are two routes. You can download the ZIP from the skill's page on our directory and unzip it into a skills folder. For a personal setup, put the unzipped openclaw-brand folder in ~/.claude/skills/. For a single project, put it in that project's .claude/skills/ folder. Alternatively, in the Claude apps, go to Settings → Capabilities → Skills and upload the ZIP there.

After installing, check that the folder contains SKILL.md at its top level, alongside references/identity.md and references/asset-rights.md. Those three files are the whole skill.

• Personal: ~/.claude/skills/openclaw-brand/
• Project: .claude/skills/openclaw-brand/ inside your project
• Claude apps: Settings → Capabilities → Skills, then upload the ZIP

How to trigger the skill with natural requests

You do not need a special command. The skill's description tells the agent when to apply it: tasks about logos, typography, imagery, voice, documents, presentations, social graphics and launch materials that change brand identity. Describe the artefact and its audience in plain language and the agent should pick the skill up.

The description also says when not to use it. Ordinary product UI and public-page composition are outside its scope, so ask for those as normal design work and expect the agent to treat them as product surfaces, not marketing pages.

• Make a 16:9 title slide for the OpenClaw launch deck using the brand palette.
• Draft a LinkedIn social graphic announcing a new release, with a sentence-case headline.
• Create a logo variant for a dark background without changing the logo's proportions.
• Write launch-post copy in the OpenClaw voice for the community newsletter.
• Check whether I can reuse the lobster artwork in a conference handout.

What the agent does when the skill is active

The agent first reads the identity reference, then identifies whether the artefact is product, documentation or marketing. For code, it uses semantic variables rather than raw hex values, and it keeps status colours for success, warning, error and information instead of using them as decoration.

Before finishing, it checks contrast, responsive cropping and text legibility. If you ask it to confirm logo clearspace, it will only do so when approved guidance defines that measurement. Otherwise it says the check is unavailable.

• Coral #f5654a (dark) or #d84a31 (light) for primary emphasis
• Sea glass #4fc8ae (dark) or #14806e (light) as a secondary accent
• Switzer or the --oc-font-display / --oc-font-body fallback stack for text
• Sentient or --oc-font-serif reserved for editorial accents and quotations
• System monospace or --oc-font-mono for code

Respect the asset-rights rules

The asset-rights reference states that this repository distributes CSS and guidance, not brand asset binaries. Logos, lobster artwork, mascot files, photos and illustrations should not be redistributed without a repository-local licence or a recorded grant. A public URL or an asset already on a website does not prove you may reuse it.

Before adding any new asset, the reference asks you to record the source and owner, the licence or written permission, allowed uses, required attribution and the repository path where that evidence lives. If any of this is unknown, the asset stays in the existing consumer and is referenced only from there. System and fallback font stacks are always acceptable. Font files for Switzer or Sentient must not be copied into a consumer or release artefact.

Safety notes: scripts and network access

According to the skill's listing, it does not run scripts and does not use the network. It has no listed requirements. It consists of Markdown instructions and reference files, so there is nothing for it to execute on your machine. Note that the skill governs how the agent designs and writes; it does not verify licences for you, so keep the asset-rights steps in your own process.

• Runs scripts: no
• Uses network: no
• Requires: nothing listed

Tips for better results

Name the audience and artefact type up front, because the skill asks the agent to classify the work as product, documentation or marketing before styling it. Say whether the output is code, so the agent uses design tokens. If you want a logo used, state where the file came from, since the asset-rights rules will otherwise block reuse.

Ask the agent to verify contrast and legibility on the final artefact, since the skill's workflow includes those checks.

• Give the audience and format first (for example, a slide or a web page)
• Ask for code output if you need tokens rather than raw colours
• Provide the asset's source and licence evidence before asking for reuse