Docs/AI tools (MCP)/Windsurf

Connect Prowlo to Windsurf

Windsurf (by Codeium) has full support for Model Context Protocol (MCP) servers. Prowlo integrates via a remote URL endpoint, allowing you to search your Dataset, browse the feed, and query live Reddit data directly within Windsurf's Cascade AI agent. No extensions or npm packages needed.

Recommended: connect by URL + OAuth

The simplest way to connect is the hosted URL https://api.prowlo.com/mcp with OAuth sign-in — no API key to copy or rotate. See the one-URL connector guide. The API-key steps below work for any client that needs a key.

Prerequisites

•Active Prowlo account
•Prowlo API key (learn how to generate an API key)
•Windsurf IDE installed (latest version recommended)

Setup Instructions

Get your API key

Open API Access in your Prowlo dashboard.

Click Generate New Key and name it something like "Windsurf". Copy the key right away — you cannot retrieve it later.

Open the Windsurf MCP configuration

You can configure MCP servers in Windsurf through either method:

Option A (File): Edit the config file directly at ~/.codeium/windsurf/mcp_config.json

Option B (Settings UI): Open Windsurf Settings, then navigate to MCP Servers and click to add a new server.

Add the Prowlo configuration

If using the file method, add this configuration to your mcp_config.json:

{
  "mcpServers": {
    "prowlo": {
      "url": "https://api.prowlo.com/mcp",
      "headers": {
        "Authorization": "Bearer prowlo_your_key_here"
      }
    }
  }
}

Replace prowlo_your_key_here with your actual API key from Step 1.

If using the Settings UI, enter the same URL and headers directly in the form provided.

Restart Windsurf and verify

Close and reopen Windsurf completely for the configuration to take effect. Open Cascade (Windsurf's AI chat) and ask: "Search my Prowlo dataset for posts about pricing". If the connection is working, Windsurf will query Prowlo and return matching records.

Tips for Using Prowlo with Windsurf

Leverage Cascade for feedback mining

Windsurf's Cascade agent excels at multi-turn conversations with code context. Ask Cascade to search your Prowlo Dataset, review your codebase, and then turn real user reports into issues or tasks. The agent can naturally blend your implementation details with the records Prowlo surfaces.

Use MCP for research loops

Ask Cascade to search your Dataset, pull live Reddit threads from Prowlo, and then synthesize what users are saying. Windsurf can maintain context across multiple queries, making it ideal for exploratory research.

Manage Watchers and tags conversationally

You can list and update the Watchers that feed your Dataset, and manage keyword tags, directly from Cascade. Ask "List my Watchers" or "Add a tag for 'API gateway'" — no need to switch to the Prowlo dashboard.

Secure your config

If you version control your Windsurf config, use environment variable substitution for the API key in production. Store the actual key in .env or your deployment secrets manager.

Troubleshooting

MCP server not appearing?

Make sure you restarted Windsurf completely after editing the config. Check the Windsurf logs (Help → Show Logs) for any JSON parsing errors or connection failures.

Authentication error?

Verify that your API key starts with prowlo_ and has been pasted correctly without extra spaces. If the key is compromised, revoke it in Prowlo and generate a new one.

Invalid JSON in config?

Use an online JSON validator to check your mcp_config.json file. Common issues include trailing commas, unmatched braces, or improper quote escaping.

Cascade not finding Prowlo tools?

After restarting, check Windsurf's MCP settings panel to confirm the Prowlo server shows as "Connected". If it still doesn't work, try disabling and re-enabling the server in settings.

← Cursor API Keys →