Quickstart

Get Kerno KIT running and connected to your coding agent in under 60 seconds.

Already using Kerno? If you have Kerno installed and your agent connected, skip to Step 4.

Prerequisites

Before you begin, make sure you have:

• Node.js 18+ installed

• Docker running on your machine

• Git initialized in your project

• A TypeScript, JavaScript, or Python project

Step 1 — Install Kerno

Install the Kerno CLI globally or run it directly with npx:

npm install -g @kerno/cli

Or without installing:

npx @kerno/cli mcp

Step 2 — Start the Kerno agent

From your project directory, run:

kerno mcp

To specify a path explicitly:

kerno mcp -w /path/to/your/project

Kerno will open a browser window to authenticate you, then print the MCP configuration for your coding tool.

Step 3 — Connect your coding agent

Add the MCP server to your coding agent using the configuration printed in the previous step. Any

Any MCP-compatible coding tool can connect using the same URL.

Claude Code:

claude mcp add --transport http kerno http://localhost:<port>/mcp

Cursor: Add to .cursor/mcp.json:

{
  "mcpServers": {
    "kerno": {
      "url": "http://localhost:<port>/mcp",
      "transport": "http"
    }
  }
}

The port is session-specific. Always copy it from your terminal output, do not assume a default.

Step 4 — Disable tools you don't need (optional)

If you are using Kerno for code intelligence only, you can disable the environment, baseline, and validation tools in your agent's MCP settings. How to disable individual tools varies by agent — refer to your agent's documentation.

The tools KIT needs:

• kerno_sync_workspace

• kerno_list_endpoints

• scan_workspace

• index_module

• package_heatmap

• symbol_heatmap

• find_usages

• find_definition

• find_references

• call_hierarchy

• list_build_files

For a full breakdown of available tools, see KIT tools reference.

Step 5 — Index your codebase

Ask your agent to scan and index your workspace:

"Use Kerno to index this project"

Kerno will scan your repository, detect modules, and build the SCIP index. The first index takes a few minutes depending on the size of your codebase. Subsequent runs reuse the existing index and update only what changed.

What's next

Your agent now has a complete map of your codebase. Try asking it:

"What does calling this function affect across the codebase?" "Find all usages of this interface." "Show me the full call chain from this method."

For a full breakdown of available tools, see Tools Reference.