AI help for content and design

Content and design work fails in a specific way: the draft looks finished, so people ship it. A poster, a voiceover, or a frame of interface can be polished and still be the wrong brand, the wrong claim, or a copy of someone else's work. The listings here are tools for making that draft, plus one skill that only knows a single company's visual identity.

Anthropic's canvas-design skill creates original posters and static art as PNG or PDF. Its instructions say to start from a short design philosophy and to avoid copying an existing artist. Anthropic's brand-guidelines skill is narrower than the name suggests. It applies Anthropic's colors and fonts, with Poppins and Lora when those fonts are installed, and Arial and Georgia as fallbacks. It will not learn your palette. If you need your own brand, you have to write those rules yourself or paste a guide you already have.

Figma's MCP server brings a Figma file into the AI app. The official guide describes generating code from a selected frame, reading variables and components, and, on the remote server, writing frames back to the canvas. Figma says write-to-canvas is free during a beta and is planned to become a paid, usage-based feature. Read tools are rate-limited: Starter plans, and View or Collab seats, are limited to six tool calls a month. Dev or Full seats on paid Figma plans follow Figma's API rate limits instead. Generated placeholder images are supposed to ask permission and disclose Figma AI credit use.

ElevenLabs' GitHub server is the company's official MCP project, and the README says the local server is deprecated. The hosted server is the one ElevenLabs tells you to use. It signs in with OAuth rather than an API key in a config file. The older README's examples cover spoken audio, voice variation, sound effects, and transcription, and it warns that ElevenLabs credits are required. A free tier with a monthly credit allotment exists, according to that README, and you can spend it by asking for audio you did not mean to generate.

Look at the artifact, not the chat. Open the PNG, play the audio, or click the Figma frame. Check names, claims, and whether the design system you already have was actually used. Code Connect, which Figma documents, is how a team keeps generated code aligned with real components. Without it, generated UI is a picture of your product, not your product.

These summaries come from the publishers' READMEs and guides in October 2026. We have not generated a poster, a voice, or a Figma frame as part of building this directory. Untested is the honest status.