Akamai Traffic MCP Server

Akamai Traffic MCP

An MCP (Model Context Protocol) server that exposes the Akamai Reporting API v2 as tools an LLM can call directly — via Claude Desktop, any MCP-compatible client, or your own agent.

Generate traffic reports by hostname or CP code, analyse HTTP error rates, measure cache offload, and forecast future traffic trends — all through natural language.

Features

Key behaviours:

• Default time window: last 15 days (adjustable per call)
• Multi-account support via accountSwitchKey — call list_accounts() first
• Forecasting requires no heavy ML dependencies — uses numpy only
• Credentials never leave your machine; all auth is via Akamai EdgeGrid

Architecture

Claude Desktop / MCP client
        │  MCP (streamable-http)
        ▼
  server.py  (FastMCP)
        │
        ├── akamai_api/client.py    EdgeGrid auth + HTTP session
        ├── akamai_api/reports.py   Request body builders
        ├── forecast.py             Linear regression & EMA forecasting
        ├── export.py               Excel report writer (openpyxl)
        └── config.py               .edgerc + .env config loader

Prerequisites

• Python 3.11+
• An Akamai ~/.edgerc file with a [default] section (or the section name of your choice)
• API credentials with access to:
  • Identity Management API v3 (for account switching)
  • Reporting API v2 (for traffic data)

Quick Start (local)

1. Clone and install

git clone https://github.com/gamittal-ak/traffic_report_mcp.git
cd traffic_report_mcp
python3.11 -m venv .venv
source .venv/bin/activate        # Windows: .venv\Scripts\activate
pip install -r requirements.txt

2. Configure environment

cp .env.example .env

Edit .env:

EDGERC_PATH=/home/you/.edgerc   # path to your .edgerc file
EDGERC_SECTION=default           # section name inside .edgerc
MCP_HOST=0.0.0.0
MCP_PORT=8000
LOG_LEVEL=INFO

3. Start the server

python server.py

The MCP server is now available at http://localhost:8000/mcp.

Connect to Claude Desktop

Edit your Claude Desktop config file:

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "akamai-traffic": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "http://localhost:8000/mcp",
        "--allow-http"
      ]
    }
  }
}

Restart Claude Desktop. The Akamai traffic tools will appear automatically.

Example conversations

"Show me traffic for all my accounts over the last 7 days"

"Which hostnames had the most edge hits in the past 15 days for account ACME Corp?"

"What were my 5xx error rates last week for CP codes 12345 and 67890?"

"Forecast my edge bytes for the next 7 days using exponential moving average"

"Export a full traffic report to Excel for account XYZ"

Tool reference

`list_accounts(search="")`

Returns all accounts accessible via your API credentials. search filters by account name (optional). Always call this first to get the accountSwitchKey needed by other tools.

→ [{accountSwitchKey, accountName, accountId}, ...]

`get_traffic_by_hostname(...)`

Traffic broken down by hostname.

Metrics returned: edgeBytesSum, edgeHitsSum, originBytesSum, originHitsSum, midgressBytesSum

`get_traffic_by_cpcode(...)`

Traffic grouped by CP code with offload percentage.

Metrics returned: edgeBytesSum, originBytesSum, midgressBytesSum, offloadedBytesPercentage

`get_http_status_breakdown(...)`

4xx and 5xx error hit counts at edge and origin. Useful for spotting error spikes and diagnosing origin health issues.

`get_edge_origin_offload(...)`

Cache offload percentage by CP code. Higher offloadedBytesPercentage = more traffic served from Akamai edge = l