MCP

Model Context Protocol
for ShadowPhone

ShadowPhone exposes a Model Context Protocol (MCP) endpoint so AI agents, Claude, and external tools can read fleet state, trigger workflows, and act on Instagram operations data through a standard protocol.

What Is MCP

A standard protocol for AI tool use

The Model Context Protocol (MCP), developed by Anthropic, is a standard for connecting AI assistants to external tools, APIs, and data sources. Any MCP-compatible client — Claude Desktop, Claude Code, or a custom agent — can connect to a server that exposes MCP endpoints and call its capabilities like native tools.

ShadowPhone implements an MCP server at /api/mcp. This means an AI agent can discover what your ShadowPhone fleet is doing, issue commands, and read results — without any custom API integration code.

MCP replaces the pattern of building one-off integrations for every AI tool. Write the connection once, and any MCP-compatible client can use it.

Claude Desktop config (claude_desktop_config.json)
{
  "mcpServers": {
    "shadowphone": {
      "url": "https://www.shadowphone.io/api/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}
Capabilities

What the MCP endpoint exposes

The ShadowPhone MCP server exposes a set of tools an AI agent can discover and call. Capabilities expand as the MCP integration matures.

Read fleet state

Query connected devices, online accounts, profile assignment, and current scheduler status across the full fleet — from any MCP-compatible AI client.

Trigger workflows

Invoke post, engagement, story, and DM workflows programmatically. The brain queues and executes them on the correct device based on account assignment.

Read analytics

Pull account-level and fleet-level metrics: reach, post count, engagement rate, and error logs — without opening the dashboard.

Manage schedules

Read and update per-slot schedules, toggle steps on or off, and inspect queue state for any profile through the MCP context layer.

Use Cases

What you can build with it

AI-directed posting

Connect Claude or another AI agent to ShadowPhone. The agent generates captions and selects content, then triggers the post workflow directly through MCP — no human copy-paste step.

Operator dashboards

Build a custom internal tool that reads live fleet state from ShadowPhone's MCP endpoint and surfaces it in your own interface alongside other data sources.

Conditional automation

Write an agent that reads account engagement metrics, decides whether to increase or decrease posting frequency, and updates the schedule — all through the MCP protocol.

Cross-platform orchestration

Use ShadowPhone's MCP alongside other MCP-compatible tools (Supabase, Airtable, email) so a single AI agent can coordinate CRM, content scheduling, and fleet execution.

Setup

Getting started

01

Get an API key

Log into your ShadowPhone account and generate an API key from the Settings → API page. The MCP endpoint requires a valid key in the Authorization header.

02

Add the server to your MCP client

For Claude Desktop, add the configuration block above to your claude_desktop_config.json. For Claude Code, use the claude mcp add command with the endpoint URL and your key.

03

Discover available tools

Your AI client will automatically discover the tools exposed by the ShadowPhone MCP server. In Claude, type / or describe what you want to do — Claude will identify and use the relevant ShadowPhone tool.

04

Test with a fleet status query

Ask your AI agent: "What devices are connected to ShadowPhone right now?" This reads the fleet state tool and confirms the connection is working before you build anything further.

Questions about the MCP integration?

The MCP endpoint is actively developed. If you need a capability that is not yet exposed, or you run into an issue connecting, reach out — operator feedback directly shapes what gets added next.