first-steps6 min

Your AI Coding Partner

One of the things that makes Clean Language genuinely different is how well it works with AI. Not just paste code and ask questions — but a real, integrated setup where the AI knows the exact language spec, can verify your code, and catches mistakes before you even run anything.

Start the MCP server from your terminal:

cln mcp-server

If you're using VS Code with the Clean Language extension, it connects automatically. Once connected, the AI looks up the current API, checks the syntax, verifies the types — and hands you back working Clean Language code.

Adding require to your functions gives the AI precise information about what they expect:

functions:
    number average(list<number> values)
        require values.length() > 0
        number total = 0.0
        iterate v in values
            total = total + v
        return total / values.length().toNumber()

start:
    list<number> scores = [8.5, 9.0, 7.5]
    print(average(scores).toString())

8.333333333333332

The require tells anyone reading — human or AI — this function assumes the list is not empty. The AI won't generate code that calls average([]) and won't suggest removing the check.

Quick recap

cln mcp-server starts the AI integration — the VS Code extension starts it automatically
AI assistants with the MCP server know the real, current Clean Language spec
Describe what you want in plain language and get back spec-correct code
require in your functions gives the AI precise expectations to work from
You always review and own the code the AI generates