AI coding agents are incredibly productive. But productive and correct aren't always the same thing.

If you've ever asked an AI assistant to integrate with Garchi CMS, you've probably experienced this firsthand. The code runs. The page renders. But the content structure is fragile, the section templates don't map cleanly to your components, and the whole setup doesn't scale beyond a demo.

That's why we're releasing Agent Skills for Garchi CMS — a structured set of best practices that teach AI agents how to render and work with your content from Garchi CMS properly.

What Are Agent Skills?

If you're new to Agent Skills, they are modular, portable folders containing instructions, metadata, and optional resources (scripts, templates) that extend AI agent capabilities in tools like Claude or GitHub Copilot. In simple words, think of agent skills as training material for your AI agent — they tell it how to do things the right way, specific to a platform.

Why Use the Garchi CMS Agent Skill?

AI agents and tools are powerful when provided with the right context and information. Creating content in Garchi CMS using the dashboard is straightforward, but what if you want tools like Claude to generate a blog article for you? Maybe you have a marketing guide for your website and you want to create content that follows the guidelines in your document. What if your AI tool could directly talk to Garchi CMS and create that content under your supervision? That's where Garchi MCP takes the lead.

So you've ticked the box of creating content in your Garchi CMS space using Garchi MCP and your AI tool. Now it's time to render it correctly and connect the bridge with your app. That's where Garchi CMS Agent Skills come into the picture. They make the job simpler for you and your AI tool because all the good practices and rules for writing content-rendering code with Garchi CMS are baked into the skill.

What's Inside

The first skill we're shipping is garchi-render-content, focused on making one of the most common developer tasks easier — rendering Garchi CMS content in your codebase correctly.

It's specific to Garchi's architecture: how sections map to components, how templates define props, and how content flows from the API to your frontend. The skill gives your AI agent the patterns that experienced Garchi developers already know, so you're not starting from scratch every time.

The skill covers the available Garchi CMS SDKs and API usage, and each pattern is structured so an AI agent can pick it up and apply it without ambiguity.

We'll be honest — it makes your life easier by teaching your AI agent "how to," but you should always verify the output against our Garchi CMS documentation and API documentation. Agents are helpful, not infallible.

How It Works With the Garchi MCP Server

This skill is designed to work alongside the Garchi MCP Server. Here's how the two fit together:

The MCP server gives your AI agent the ability to interact with Garchi CMS — creating pages, managing sections, publishing content, handling metadata, and working with assets. It's the hands.

The agent skill gives your AI agent the knowledge to render that content correctly in your codebase. It's the brain.

You need both. The MCP server lets your agent say "create a landing page with a hero section and three feature cards." The agent skill ensures that when it generates the frontend code to render that page, it follows the right patterns — dynamic component mapping, proper prop handling, correct API fetching, and draft mode awareness.

Without the skill, the agent will get creative. And creative, in this context, usually means inconsistent.

Setting It Up

Step 1: Install the Garchi MCP Server

First, make sure your AI client is connected to Garchi through the MCP server. We support:

• Cursor IDE
• Claude Desktop
• Claude CLI
• OpenAI Codex
• Lovable
• v0
• VS Code Copilot
• Other clients

Step 2: Add the Agent Skill

Install with the Skills CLI:

npx skills add lumenharbor/garchi-cms-skills

Using Claude Desktop?

You can install it as a plugin directly:

1. Open Claude Desktop and find the Cowork tab
2. Click on Plugins from the left sidebar
3. In the Browse Plugins popup, select Add marketplace from GitHub
4. Enter lumenharbor/garchi-cms-skills as the URL and click Sync

Step 3: Use It

Once installed, invoke the skill manually by typing /garchi-render-content in your input followed by your prompt, or let your AI agent pick up the skill automatically. For example:

/garchi-render-content Build a Next.js page that renders my Garchi landing page sections as React components

The agent will reference the skill's rendering patterns and generate code that follows Garchi CMS best practices — proper section-to-component mapping, correct SDK usage, draft/live mode handling, and consistent data fetching.

Documentation You Should Know

The agent skill works best when you're also familiar with the Garchi ecosystem:

• Garchi API Documentation (v2) — The full API reference
• Garchi CMS Documentation — Platform guides and concepts
• MCP Documentation — Setting up AI client integrations

The skill itself references these docs internally, so your agent has the context it needs. But it's worth bookmarking them for your own reference too — agents can make mistakes, and having the docs handy means you can verify when something looks off.

What's Next

This is v1.0.0, and the garchi-render-content skill is just the beginning. We're actively expanding based on real-world usage. SEO optimisation for headless setups, advanced section nesting — these are all on the roadmap.

If you're building with Garchi CMS and your AI agent keeps making the same mistake, that's a rule we should add. Open an issue on the GitHub repo or reach out directly.

The goal is simple: your life with Garchi CMS, made easier.