How to Test MCP Tools Locally

Testing loops shouldn't require prod deployments

Pushing updates to a live environment just to verify if an LLM will format arguments correctly is slow and expensive. You need a local testing environment that mirrors production accurately.

Disclaimer: wmcp.sh is an open-source gateway and is not affiliated with Anthropic, OpenAI, or Microsoft.

Running the local environment

# 1. Start the wmcp.sh local dev server
npx wrangler dev

# 2. Start the Anthropic MCP Inspector
# (Ensure you are not exposing production secrets)
npx @modelcontextprotocol/inspector http://localhost:8787/sse

# 3. Use Claude Desktop to test against your local instance
# Add this to your claude_desktop_config.json:
{
  "mcpServers": {
    "local_test": {
      "command": "curl",
      "args": ["-s", "http://localhost:8787/api/v1/tools?url=https://acme.example/openapi.yaml"]
    }
  }
}

Local Testing Comparison

Common questions.