Skip to main content
Glama
Sign In
analytics-mcp

analytics-mcp

by bhavya998
OverviewSchemaSourceRelated ServersDiscussionsAdmin
Python
Hybrid

analytics-mcp

Production-grade MCP server for enterprise sales analytics. 6 tools, 3 resources, 2 prompts — all with structured outputs — over a SQLite database. Connects to Claude, OpenCode, Cursor, and any MCP client.

Python FastMCP Tests License

What This Is

A Model Context Protocol (MCP) server that exposes an enterprise sales database to LLM clients. Instead of copy-pasting data into ChatGPT, connect this server and let the AI query, analyze, and visualize your data through structured tools.

Built with the latest MCP patterns (July 2026):

  • FastMCP 3.4 — high-level server framework

  • Structured outputs — Pydantic models define outputSchema for every tool

  • Tool annotations — readOnlyHint lets clients skip confirmations

  • In-memory client testing — no port flakiness, era-neutral

  • Resources — schema introspection, table details, report templates

  • Prompts — reusable analysis templates (sales analysis, customer segmentation)

  • Both transports — stdio (local) + HTTP (remote)

Related MCP server: agente-ecommerce-mcp-memoria

Quick Start

git clone https://github.com/bhavya998/analytics-mcp.git
cd analytics-mcp
uv sync

# Initialize database (200 customers, 25 products, 2000 orders)
uv run analytics-mcp init

# Start server (stdio for local MCP clients)
uv run analytics-mcp serve

# Or HTTP transport for remote access
uv run analytics-mcp serve --transport http --port 8000

Connect to Claude Desktop / OpenCode

Add to your MCP client config:

{
  "mcpServers": {
    "analytics": {
      "command": "uv",
      "args": ["run", "--directory", "/path/to/analytics-mcp", "analytics-mcp", "serve"]
    }
  }
}

Then ask: "Show me the top 5 products by revenue and analyze the monthly sales trend"

Tools

Tool

Description

Annotations

query_database

Run parameterized SELECT queries against the database

readOnlyHint

get_revenue_report

Revenue by category/region/tier/status/month with profit analysis

readOnlyHint

get_customer_profile

360-degree customer view: orders, LTV, favorite category, recent activity

readOnlyHint

get_top_products

Leaderboard by revenue or quantity, optional category filter

readOnlyHint

analyze_sales_trend

Time-series with period-over-period growth rates

readOnlyHint

get_regional_performance

Ranked regional performance with revenue, orders, customers, AOV

readOnlyHint

Resources

URI

Description

schema://database

Full database schema (all tables, columns, types, row counts)

schema://tables/{table_name}

Detailed table schema with sample rows

report://templates

Available report templates and usage examples

Prompts

Prompt

Description

sales_analysis

Comprehensive sales analysis with focus area (overall/category/region/customer)

customer_segmentation

Segment customers into Champions/At Risk/New/Dormant with retention strategies

Database Schema

customers (200 rows)
  id, name, email, company, region, tier, signup_date, lifetime_value

products (25 rows)
  id, name, category, price, cost, stock

orders (2000 rows)
  id, customer_id, employee_id, order_date, status, total

order_items (5000+ rows)
  id, order_id, product_id, quantity, unit_price

employees (15 rows)
  id, name, role, region, hire_date

Testing

make test         # 43 tests: database, tools, full MCP client integration
make lint         # ruff

Suite

Tests

Pattern

test_database.py

8

DB init, query validation, schema introspection

test_tools.py

17

Direct tool function calls (all 6 tools)

test_server_client.py

12

Full MCP protocol via in-memory Client

Tests use the modern in-memory Client(mcp) pattern — no ports, no subprocesses, era-neutral.

Tech Stack

Layer

Technology

MCP Framework

FastMCP 3.4 (Prefect)

Protocol

MCP 1.28 (Streamable HTTP + stdio)

Database

SQLite with seeded enterprise data

Validation

Pydantic v2 (structured tool outputs)

CLI

Typer + Rich

Testing

pytest + pytest-asyncio + FastMCP Client

Project Structure

analytics-mcp/
├── src/analytics_mcp/
│   ├── server.py         MCP server: 6 tools, 3 resources, 2 prompts
│   ├── database.py       SQLite setup + schema + seed data (2000+ orders)
│   ├── schemas.py        Pydantic models for structured outputs
│   └── cli.py            CLI (serve, init, inspect)
├── tests/                43 tests (database, tools, client integration)
├── data/                 SQLite database (auto-generated, gitignored)
├── Makefile
└── pyproject.toml

License

MIT

Available Tools

6 tools
analyze_sales_trendA
Read-only

Analyze sales trends over time with period-over-period growth rates.

Returns revenue, order count, and growth percentage for each period.

Inputs: period_end, granularity, period_start. Returns: Sales trend over time.

get_customer_profileA
Read-only

Get a 360-degree customer view: profile, order history, spending stats, favorite category.

Useful for account reviews, churn analysis, and upsell identification.

Inputs: customer_id (required). Returns: 360-degree customer view.

get_regional_performanceA
Read-only

Get ranked sales performance by region: revenue, orders, customers, AOV.

Returns all 5 regions ranked by total revenue.

Inputs: no named parameters. Returns: Ranked regional performance report.

get_revenue_reportA
Read-only

Generate a revenue report grouped by a dimension within a date range.

Returns total revenue, order count, average order value, and estimated profit per segment.

Inputs: group_by, period_end, period_start. Returns: Full revenue report across a dimension.

get_top_productsA
Read-only

Get a leaderboard of top-performing products by revenue or quantity sold.

Optionally filter by category.

Inputs: limit, metric, category. Returns: Top products leaderboard.

query_databaseA
Read-only

Run a read-only SQL query against the enterprise database.

Tables: customers, products, orders, order_items, employees.

Examples: SELECT * FROM customers WHERE tier = 'enterprise' LIMIT 10 SELECT region, COUNT(*) as customers FROM customers GROUP BY region SELECT p.name, SUM(oi.quantity) as sold FROM order_items oi JOIN products p ON oi.product_id = p.id GROUP BY p.name ORDER BY sold DESC LIMIT 5

Inputs: sql (required). Returns: Result of a SQL query.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    Deploy Server
    One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

    TDQS

    A3.9/5.0

    Scored across 6 tools

    Disambiguation4/5

    Tools are mostly distinct: each targets a specific analytics aspect (trends, customer profile, regional, revenue report, top products, raw SQL). Some overlap in subject matter but purposes and outputs are clearly differentiated.

    Naming Consistency3/5

    Most tools use 'get_' prefix, but 'analyze_sales_trend' and 'query_database' break the pattern. The naming is readable but inconsistent across verbs.

    Tool Count5/5

    6 tools is well-scoped for an analytics server, covering common analytical functions without being excessive. Each tool earns its place.

    Completeness4/5

    Covers key analytics areas: trends, customers, regions, revenue, products. The query_database tool compensates for most gaps. Minor omission like employee analytics is noted but not critical.

    Resources

    Related MCP Connectors

    Related MCP Servers

    • agente-ecommerce-mcp-memoria

      AbelCierto
      F
      license
      Not graded
      quality
      D
      maintenance
      An MCP server that exposes a LangChain agent with short-term memory to analyze real e-commerce data through predefined SQL tools, enabling natural language queries on sales, customers, and logistics.
      -