Mcpspan

Mcpspan: Self-hosted analytics software that monitors tool calls, response times, failures, and token usage for Model Context Protocol servers.

Mcpspan provides open-source, self-hosted analytics for Model Context Protocol (MCP) servers. Developers integrate it by adding one line of code to their server and running a Docker Compose dashboard locally. The platform tracks tool calls, resource reads, and prompt gets from clients including Claude, Claude Code, Cursor, VS Code, Codex, and ChatGPT. It reports median and 95th percentile latencies, response payload sizes, schema token costs, and detailed failure categories. SDKs are available across eight languages: TypeScript, Python, Go, C#/.NET, Java/Kotlin, Rust, Ruby, and PHP. Data remains local, with optional exports to CSV, NDJSON, or OpenTelemetry endpoints.

Mcpspan at a glance

Best for: Software engineer
Runs on: self-hosted, open source, VS Code, Slack
Pricing: Free
Released: October 2026
Category: AI coding tool
Website: github.com/mcpspan/mcpspan

Key features

Tracks tool calls, resource reads, and prompts per client
Measures median, 95th percentile, and full response time distributions
Monitors typical and maximum tool response sizes
Calculates token costs of tool definition lists shown to agents
Categorizes server exceptions, rejected arguments, and missing tools
Traces step-by-step session calls and detects agent repetition loops
Marks server release boundaries and schema changes on usage charts
Sends error and inactivity alerts to Slack, Discord, or custom webhooks

Use it for

Identifying unused tools that inflate prompt context and token costs
Debugging agent execution loops and tool argument validation errors
Comparing MCP server latency and error rates across releases
Auditing MCP client requests from Claude, Cursor, and VS Code
Exporting MCP telemetry data to Grafana, Datadog, or Honeycomb

Alternatives to Mcpspan

MCP Inspector — Test and debug MCP servers via web, CLI, and terminal interfaces.
Coding Tools MCP — Turns any chat app into a coding agent with strict workspace safety boundaries.
MCPX — Monetize Model Context Protocol servers and build AI agent tools.
MCP Toolbox for Databases — Connects AI agents to diverse enterprise databases with built-in security and observability.
Finance Toolkit MCP Server — Calculates over 500 financial methods through code instead of model memory.
jCodeMunch MCP — Cuts AI code-reading token costs by over 95 percent via tree-sitter symbol indexing.
MetaMCP — Centralizes MCP server management with enterprise authentication and scaling.
Respan — Route requests to over 1,000 models with built-in fallback and observability.
Codebase Memory MCP — High-speed indexing and persistent knowledge graphs for AI coding agents.
AWS MCP Servers — Provides real-time access to official AWS documentation and infrastructure tools.
Spec Workflow MCP — It brings sequential spec creation and real-time tracking to your AI workflow.
mcpgawk — Pins each tool to what you approved and blocks calls if a server changes.

All Mcpspan alternatives