Supercharging Your Coding Assistant with Outflo's MCP Server

Model Context Protocol (MCP) is a game-changer for AI-assisted engineering and automation. Developed by Anthropic, MCP provides a standard protocol for AI models to safely and securely interact with external tools, databases, and APIs.

By connecting your favorite AI coding assistant (such as OpenAI Codex, Cursor, Windsurf, or VS Code extensions like Cline and Roo Code) to the Outflo MCP Server, you can command your LinkedIn outreach campaigns, inspect lead lists, check conversations, and register connections directly from your agentic workspace.

This guide covers how to configure the integration for your coding assistant using our hosted Server-Sent Events (SSE) gateway.

1. Outflo MCP Server Capabilities

Once connected, your AI coding assistant gains native access to Outflo's public API tools. Instead of copy-pasting JSON data or manually navigating the dashboard, you can simply ask your AI assistant to:

List and Search Campaigns: Ask "How are my active campaigns performing?" to get summaries, status, and lead volumes.
Retrieve Campaign Leads: Ask "Show me the last 15 leads from our CEO Outreach campaign" to inspect specific responses and profiles.
Manage Lead Statuses: Programmatically pause or resume campaign outreach for individual leads when they reply.
Add New Leads: Import fresh LinkedIn profile URLs or custom lead payloads directly into an active sequence.

2. Connecting via OpenAI Codex (OAuth/OIDC)

For cloud-based AI tools like OpenAI Codex, the simplest connection method is our hosted Server-Sent Events (SSE) gateway. This gateway secures your Outflo workspace using standard OAuth2 authentication.

Step 1: Add the MCP Server

To configure the hosted production or development servers in Codex, add the endpoints under `mcp_servers` in your global config file at `~/.codex/config.toml`:

Code Snippet

[mcp_servers.outflo-prod]
url = "https://mcp.outflo.io/mcp"

[mcp_servers.outflo-dev]
url = "https://mcp-dev.outflo.io/mcp"

Step 2: Log In

Once the configuration is saved, run the login flow in your terminal:

Terminal Command

codex mcp login outflo-prod

This opens your web browser to authenticate with your Outflo credentials. Once authorized, Codex stores the OIDC token and establishes a secure connection to the MCP server.

3. Connecting via Cursor / Windsurf / VS Code Extensions

For desktop assistants that support Server-Sent Events (SSE), you can connect to the hosted server directly.

If the assistant does not support OAuth browser redirection flows, you can authenticate using your Outflo Personal Access Token (API Key) as a Bearer token or authorization header.

Step 1: Generate an API Key

Step 2: Add the SSE Server

In your editor's MCP configuration settings:

Add a new server with type SSE.
URL: `https://mcp.outflo.io/mcp`
Headers: Add an `Authorization` header with the value `Bearer YOUR_PERSONAL_ACCESS_TOKEN_HERE` (containing your token directly).

4. Example Prompts to Try

Once configured, restart your coding assistant or run a status check to verify the connection. You can then interact with Outflo naturally:

User: Check my Outflo status and list my top 3 active campaigns. Assistant: Connecting to Outflo... You are logged into Workspace ID `35ed7acd`. Here are your top 3 active campaigns: 1. Founder Outreach (320 leads, 45% reply rate) 2. US Tech CEOs (140 leads, 22% reply rate) 3. Inbound Warmups (50 leads, 18% reply rate)

User: Add this lead to the 'US Tech CEOs' campaign: Name: Amit Kumar, LinkedIn: https://www.linkedin.com/in/amit-kumar Assistant: Calling tool `outflo.add_lead_to_campaign`... Success! Amit Kumar has been successfully imported and added to sequence step 1.

User: We just closed a deal with Neeraj Singla. Pause his outreach campaign. Assistant: Calling tool `outflo.pause_leads`... Successfully paused outreach for Neeraj Singla in campaign `US Tech CEOs`.