# Which AI Coding Assistants Are Compatible with Hallmark? > Discover which AI coding assistants Hallmark supports. Hallmark works with any assistant that uses skill or tool calling, like ChatGPT, Claude, Gemini, and Together AI. - Repository: [Hassan El Mghari/hallmark](https://github.com/Nutlope/hallmark) - Tags: getting-started - Published: 2026-08-15 --- **Hallmark is compatible with any AI coding assistant that supports skill or tool calling, including ChatGPT, Claude, Gemini, and Together AI, because it exposes a generic skill interface defined in [`skills/hallmark/SKILL.md`](https://github.com/Nutlope/hallmark/blob/main/skills/hallmark/SKILL.md).** Hallmark is an open-source design skill that encodes UI best practices for AI coding assistants. As implemented in the Nutlope/hallmark repository, the system deliberately avoids hard-coding integrations to remain assistant-agnostic, allowing any LLM-driven coding assistant to invoke its design guidelines through standard skill invocation protocols. ## How Hallmark Integrates with AI Coding Assistants Hallmark operates as a **design skill** rather than a standalone application. According to [`skills/hallmark/SKILL.md`](https://github.com/Nutlope/hallmark/blob/main/skills/hallmark/SKILL.md), the project defines itself as: > "A design skill for **AI coding assistants**. Makes the UIs they generate look made, not generated." Because it exposes a standard skill interface, Hallmark accepts JSON requests from any assistant capable of skill invocation. There is no vendor-specific API integration; instead, the assistant sends a skill call to Hallmark's endpoint and receives design system guidance in return. ## Verified Compatible AI Coding Assistants While Hallmark theoretically supports any skill-capable assistant, the repository specifically notes testing and compatibility with the following AI coding assistants: ### ChatGPT and OpenAI Models ChatGPT integrates with Hallmark via the generic skill-calling interface. Whether invoked through a ChatGPT plugin or a custom tool implementation, Hallmark processes requests from OpenAI models the same way as any other assistant. ### Claude (Anthropic) Claude's tool-calling capabilities can reference the Hallmark skill definition. When Claude issues a tool call to Hallmark's endpoint, the system returns design guidelines without requiring Anthropic-specific modifications. ### Gemini (Google AI) Google's Gemini models can invoke Hallmark through their function-calling or tool-calling mechanisms. The skill interface remains consistent regardless of whether the request originates from Gemini or other assistants. ### Together AI Hallmark's runtime infrastructure is **powered by Together AI**, providing native integration with their model endpoints. The [`package.json`](https://github.com/Nutlope/hallmark/blob/main/package.json) file explicitly states: ```json { "description": "A design skill for AI coding assistants. Makes the UIs they generate look made, not generated. Powered by Together AI." } ``` This infrastructure choice ensures optimal performance with Together AI while maintaining compatibility with other providers. ## Technical Architecture The assistant-agnostic design stems from Hallmark's skill definition structure. The [`skills/hallmark/SKILL.md`](https://github.com/Nutlope/hallmark/blob/main/skills/hallmark/SKILL.md) file encodes what the repository calls the "anti-slop consensus" for UI design. Additionally, the [`site/index.html`](https://github.com/Nutlope/hallmark/blob/main/site/index.html) meta tags reinforce this positioning: > "A design skill for **AI coding assistants**. Encodes the anti-slop consensus... Powered by Together AI." This architecture means future AI coding assistants—whether from existing vendors or new entrants—can integrate with Hallmark immediately if they support standard skill or tool invocation protocols. ## Summary - Hallmark exposes a **generic skill interface** via [`skills/hallmark/SKILL.md`](https://github.com/Nutlope/hallmark/blob/main/skills/hallmark/SKILL.md), making it compatible with any AI coding assistant that supports skill calls - The repository verifies compatibility with **ChatGPT/OpenAI**, **Claude**, **Gemini**, and **Together AI** - Hallmark's runtime is **powered by Together AI**, providing native integration with their model endpoints - No hard-coded integrations exist; assistants invoke Hallmark by sending JSON requests to its skill endpoint - Future AI coding assistants can integrate immediately if they support tool or function calling ## Frequently Asked Questions ### Does Hallmark require specific API integrations for each AI coding assistant? No. Hallmark uses a generic skill definition that any AI coding assistant can invoke through standard tool-calling or skill-calling protocols. As long as the assistant can send a JSON request to Hallmark's endpoint, it can access the design system guidance without vendor-specific code. ### Is Hallmark limited to the assistants listed in the documentation? No. While the Nutlope/hallmark repository specifically mentions testing with ChatGPT, Claude, Gemini, and Together AI, the skill architecture is deliberately assistant-agnostic. Any current or future AI coding assistant that supports skill invocation can use Hallmark. ### Why does Hallmark use Together AI specifically? According to [`package.json`](https://github.com/Nutlope/hallmark/blob/main/package.json) and [`site/index.html`](https://github.com/Nutlope/hallmark/blob/main/site/index.html), Hallmark is "Powered by Together AI" for its runtime infrastructure. This means Together AI hosts the model endpoints that power Hallmark's responses, though the skill itself remains accessible to assistants from other providers like OpenAI and Anthropic. ### What is a "design skill" in the context of AI coding assistants? A design skill is a structured set of guidelines and capabilities that AI coding assistants can invoke to improve specific outputs—in this case, user interface design. Hallmark's [`SKILL.md`](https://github.com/Nutlope/hallmark/blob/main/SKILL.md) definition encodes UI best practices that assistants reference when generating code, helping them create interfaces that look "made, not generated."