Graylog MCP Server
Enables AI assistants to query and analyze logs from Graylog instances using universal search with relative or absolute time windows, supporting both full result retrieval and lightweight count-only queries.
README
Graylog MCP Server
Introduction
The Graylog MCP Server lets AI IDEs and agents securely query your Graylog instance via the Model Context Protocol. It exposes standardized tools so assistants can search recent or absolute time windows and optionally count results without pulling full payloads.
What you get:
- search tools for Graylog universal search
- relative window: last N seconds
- absolute window: explicit ISO timestamps
- count-only variants for lightweight analytics
- drop-in configuration for popular IDEs and MCP tools
Requirements:
- a reachable Graylog URL
- credentials with permissions to use Universal Search
Links:
- Model Context Protocol: https://modelcontextprotocol.io
- Graylog: https://www.graylog.org/
Installation and Usage
Quick start (runs the MCP server over stdio):
npx -y graylog-mcp
Required environment variables:
- GRAYLOG_BASE_URL: your Graylog base URL (e.g., https://graylog.example/)
- GRAYLOG_USERNAME: Graylog username
- GRAYLOG_PASSWORD: Graylog password
Configure in your IDE or Agentic Tool of choice (Cursor, VS Code, Claude Code):
{
"graylog": {
"command": "npx -y graylog-mcp",
"env": {
"GRAYLOG_BASE_URL": "https://YOUR_GRAYLOG_INSTANCE_URL/",
"GRAYLOG_USERNAME": "YOUR_USERNAME",
"GRAYLOG_PASSWORD": "YOUR_PASSWORD"
}
}
}
Sample Usage Prompts
Some sample prompts to make the most of the MCP server:
Analyzing error patterns
search graylog for the errors in the past 24 hours with log_level:ERROR with a max limit of 100 per query
use the message patterns in the query results to figure out the patterns of errors that are occuring and put them in ERRORS.md
for subsequent queries, use the NOT condition to filter out messages with error patterns that are already discovered
Security notes:
- Prefer scoped, least-privilege Graylog credentials.
- Do not commit secrets to source control; use environment managers where possible.
Contribution and Local Development
Prerequisites:
- Bun: https://bun.sh/
- Node-compatible environment
Install and build:
# Install deps (if any) and build
bun install
bun run build
Run locally (TypeScript directly via Bun stdio):
# Start the MCP server from source
export GRAYLOG_BASE_URL="https://your-graylog.example/"
export GRAYLOG_USERNAME="your-user"
export GRAYLOG_PASSWORD="your-password"
bun index.ts
Test against a live Graylog (verifies universal search endpoints):
export GRAYLOG_BASE_URL="https://your-graylog.example/"
export GRAYLOG_USERNAME="your-user"
export GRAYLOG_PASSWORD="your-password"
# Run verification (Bun executes TypeScript directly)
bun run test:graylog
Project scripts:
- build:
bun run build→ emitsdist/index.js - test:
bun run test:graylog→ health checks for relative/absolute universal search - show-package-name: prints the package name
Debug with MCP Inspector against local source:
npx -y @modelcontextprotocol/inspector "bun index.ts"
Code style and contributions:
- Keep code readable and well-typed; avoid unnecessary complexity.
- Match existing formatting; keep lines reasonably wrapped.
- Open issues/PRs with clear reproduction steps or proposed changes.
推荐服务器
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。
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。
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。
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。
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。
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。
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。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
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。