pyroscope-mcp

pyroscope-mcp

A read-only MCP server for Grafana Pyroscope that lets AI assistants query continuous profiling data, including flamegraphs, hotspot functions, memory allocations, and label discovery, directly from any MCP-compatible client.

Category
访问服务器

README

pyroscope-mcp

A read-only Model Context Protocol (MCP) server for Grafana Pyroscope. Lets AI assistants query continuous profiling data — flamegraphs, hotspot functions, memory allocations, label discovery — directly from any MCP-compatible client.

Note: This server is query-only. Profile ingestion is expected to be handled by your application profilers (Pyroscope SDKs, agents, or exporters).

Tools

Tool Description
pyroscope_render_profile Flamegraph + timeline data from /pyroscope/render
pyroscope_label_names List all label names in a time range
pyroscope_label_values List values for a given label (e.g. service_name)
pyroscope_profile_types List available profile types (cpu, memory, goroutines…)
pyroscope_series Fetch matching label sets for a selector
pyroscope_connect_query Raw access to any /querier.v1.QuerierService/* endpoint

Prerequisites

  • Node.js 20+
  • A running Pyroscope instance (local, self-hosted, or Grafana Cloud)

Setup

git clone https://github.com/your-org/pyroscope-mcp
cd pyroscope-mcp
npm install
npm run build

Configuration

All settings are provided via environment variables:

Variable Default Description
PYROSCOPE_BASE_URL http://localhost:4040 Pyroscope server URL
PYROSCOPE_AUTH_TOKEN Bearer token (Grafana Cloud or Azure AD)
PYROSCOPE_TENANT_ID Multi-tenant org ID (X-Scope-OrgID header)
PYROSCOPE_TIMEOUT_MS 30000 Request timeout in milliseconds

Copy .env.example to .env and fill in your values:

cp .env.example .env

Adding to MCP Clients

VS Code (GitHub Copilot)

Create or edit .vscode/mcp.json in your workspace:

{
  "servers": {
    "pyroscope": {
      "type": "stdio",
      "command": "node",
      "args": ["${workspaceFolder}/dist/index.js"],
      "env": {
        "PYROSCOPE_BASE_URL": "http://localhost:4040"
      }
    }
  }
}

Then open the MCP: List Servers command in VS Code (Cmd+Shift+P) and start the server. Copilot Chat will automatically discover the tools.

For a global (user-level) config instead of per-workspace, add the same block to your VS Code settings.json under "mcp":

{
  "mcp": {
    "servers": {
      "pyroscope": {
        "type": "stdio",
        "command": "node",
        "args": ["/absolute/path/to/pyroscope-mcp/dist/index.js"],
        "env": {
          "PYROSCOPE_BASE_URL": "http://localhost:4040"
        }
      }
    }
  }
}

Claude Desktop

Edit ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):

{
  "mcpServers": {
    "pyroscope": {
      "command": "node",
      "args": ["/absolute/path/to/pyroscope-mcp/dist/index.js"],
      "env": {
        "PYROSCOPE_BASE_URL": "http://localhost:4040",
        "PYROSCOPE_AUTH_TOKEN": ""
      }
    }
  }
}

Restart Claude Desktop. A hammer icon will appear in the chat input when the server is active.

On Windows, wrap the command:

{
  "mcpServers": {
    "pyroscope": {
      "command": "cmd",
      "args": ["/c", "node", "C:\\path\\to\\pyroscope-mcp\\dist\\index.js"],
      "env": {
        "PYROSCOPE_BASE_URL": "http://localhost:4040"
      }
    }
  }
}

Cursor

Open Cursor Settings → MCP and add a new server entry, or edit ~/.cursor/mcp.json:

{
  "mcpServers": {
    "pyroscope": {
      "command": "node",
      "args": ["/absolute/path/to/pyroscope-mcp/dist/index.js"],
      "env": {
        "PYROSCOPE_BASE_URL": "http://localhost:4040"
      }
    }
  }
}

Reload Cursor. The tools appear automatically in the Composer agent context.


Zed

Add to your Zed settings.json (open via Zed → Settings):

{
  "context_servers": {
    "pyroscope": {
      "command": {
        "path": "node",
        "args": ["/absolute/path/to/pyroscope-mcp/dist/index.js"],
        "env": {
          "PYROSCOPE_BASE_URL": "http://localhost:4040"
        }
      }
    }
  }
}

Any MCP-compatible client (generic stdio config)

{
  "mcpServers": {
    "pyroscope": {
      "command": "node",
      "args": ["/absolute/path/to/pyroscope-mcp/dist/index.js"],
      "env": {
        "PYROSCOPE_BASE_URL": "http://localhost:4040",
        "PYROSCOPE_AUTH_TOKEN": "your-bearer-token-if-needed",
        "PYROSCOPE_TENANT_ID": "your-org-id-if-needed"
      }
    }
  }
}

Example prompts once connected

  • "List all services sending profiles in the last hour."
  • "Show the hottest CPU functions for media-agent in the last 30 minutes."
  • "What is the memory allocation hotspot in api-gateway?"
  • "Compare CPU usage between worker-service and data-pipeline."
  • "What profile types are available for checkout-api?"

Notes

  • For multi-tenant Pyroscope, set PYROSCOPE_TENANT_ID or pass tenantId per tool call.
  • pyroscope_connect_query gives raw access to advanced endpoints:
    • /querier.v1.QuerierService/SelectMergeStacktraces
    • /querier.v1.QuerierService/SelectSeries
    • /querier.v1.QuerierService/Diff
  • For Grafana Cloud Pyroscope, set PYROSCOPE_BASE_URL to your stack URL and PYROSCOPE_AUTH_TOKEN to a service account token.

Query smoke test

  1. Start Pyroscope locally:
docker run -d --name pyroscope -p 4040:4040 grafana/pyroscope:latest
  1. Query label names from the last hour:
NOW_MS=$(($(date +%s)*1000))
START_MS=$((NOW_MS-3600000))
curl -sS -H 'Content-Type: application/json' \
  -d "{\"start\":$START_MS,\"end\":$NOW_MS}" \
  http://localhost:4040/querier.v1.QuerierService/LabelNames
  1. Query service names from the last hour:
NOW_MS=$(($(date +%s)*1000))
START_MS=$((NOW_MS-3600000))
curl -sS -H 'Content-Type: application/json' \
  -d "{\"name\":\"service_name\",\"start\":$START_MS,\"end\":$NOW_MS}" \
  http://localhost:4040/querier.v1.QuerierService/LabelValues

推荐服务器

Baidu Map

Baidu Map

百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。

官方
精选
JavaScript
Playwright MCP Server

Playwright MCP Server

一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。

官方
精选
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。

官方
精选
本地
TypeScript
VeyraX

VeyraX

一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。

官方
精选
本地
graphlit-mcp-server

graphlit-mcp-server

模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。

官方
精选
TypeScript
Kagi MCP Server

Kagi MCP Server

一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

Exa MCP Server

模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。

官方
精选