elmah.io MCP 远程 MCP 服务器是什么?
elmah.io MCP 的远程 MCP 服务器是一个托管的 Model Context Protocol 端点,地址是 https://mcp.elmah.io/mcp,AI 助手无需在本地安装或运行任何东西即可连接。

@elmah.io
基本信息
The elmah.io MCP server allows AI assistants to interact directly with your error logs, deployments, and organization statistics. Debug production issues faster than ever before.
https://mcp.elmah.io/mcpclaude mcp add elmah-io-mcp --transport http https://mcp.elmah.io/mcp工具列表暂未提供
当前收录暂时没有可展示的工具列表,不代表服务没有工具。请查阅项目文档了解支持的能力。
此服务需要认证,获取工具列表可能需要登录或使用服务提供方的凭据。
点击上方按钮,直接连接服务器端点实时获取工具列表。
Error logging and uptime monitoring for AI assistants. Connect your AI tools directly to elmah.io to search logs, investigate errors, check deployments, and monitor uptime, all through natural language.
Server URL: https://mcp.elmah.io/mcp
Transport: Streamable HTTP
Auth: OAuth2 (interactive clients) or API key Bearer token (scripts, CI/CD)
Status: Beta
elmah.io is a cloud-based error logging and uptime monitoring service for .NET and other platforms. Applications send errors and heartbeats to elmah.io, where they're searchable, groupable, and alertable. The MCP server gives AI assistants read access to that data, plus a few targeted write actions (hide, fix, reopen, and mark messages as bot traffic), so you can debug production issues without leaving your editor or chat client.
Ask "show me the 5 most recent errors in my Production log" or "what changed since my last deployment," and your assistant pulls the answer straight from live data instead of you digging through a dashboard. It can fetch a full stack trace by ID, so there's no copy-pasting between tabs, and group errors by frequency to surface which bugs are actually worth fixing. Deployments and uptime checks are queryable the same way: list recent releases with their error counts, or check whether a monitored endpoint is currently up. Authenticate with an API key instead of OAuth and every call runs against your organization's full set of logs, not just what one user account can see.
| Tool | Description |
|---|---|
users_get_current | Fetch the current user details. |
system_get_context | Get system information like current date and time. Useful when calculating time intervals. |
organizations_list | List all organizations the current user belongs to (or the current organization, when authenticated with an API key). |
organizations_get_details | Get detailed organization info: configuration, subscription, and usage statistics. |
logs_list | List logs the current user has access to. |
logs_get_details | Get detailed configuration and metadata for a specific log. |
messages_get | Fetch a single log message by ID, including stack trace and metadata. |
messages_list_recent | List the most recent log messages for a log. |
messages_list_frequent | Find the most frequent/common error groups in a log, useful for spotting noisy bugs. |
messages_count | Count log messages by severity and timeframe. |
messages_hide / messages_unhide | Hide or unhide a log message by ID (optionally, every message sharing its grouping hash). |
messages_fix / messages_open | Mark a log message as fixed, or reopen it. |
messages_bot / messages_unbot | Mark or unmark log messages as bot-generated traffic. |
deployments_list | List recent deployments for a log, optionally including error metrics per deployment. |
uptime_list | List uptime checks for a log with current status and 24h up/down percentage. |
uptime_get_details | Get real-time results and regional performance metrics for an uptime check. |
heartbeats_list | List heartbeats configured for a log with current status. |
heartbeats_get_details | Get detailed check-in history for a heartbeat. |
recommendations_list | List automated recommendations for improving error monitoring setup, security, and configuration. |
recommendations_get_details | Get the full description and suggestions for a single recommendation. |
Destructive tools (hide, fix, reopen, unhide, bot marking) only run when a user explicitly asks. The tool descriptions themselves instruct the assistant not to call them speculatively while just investigating.
Add the server to your client using the streamable HTTP URL:
Most clients (Claude Desktop, Claude Code, VS Code, Cursor, Visual Studio, ChatGPT) walk you through an OAuth2 sign-in the first time you connect. Tokens are valid for 30 days.
Claude Code, from the terminal:
claude mcp add --transport http --client-id claudecode elmahio https://mcp.elmah.io/mcp
Then run /mcp inside Claude Code and authenticate.
Scripted / CI access: pass an elmah.io API key as a Bearer token instead of going through OAuth. Requests then run in the context of your organization rather than a specific user.
curl -s -X POST https://mcp.elmah.io/mcp \
-H "Authorization: Bearer API_KEY" \
-H "Content-Type: application/json" \
-H "Accept: application/json, text/event-stream" \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"logs_list","arguments":{}}}'
Full setup guides for each client: https://docs.elmah.io/setup-mcp-server/
elmah.io AI Plugin
For Claude Code and Cursor, the elmah.io AI Plugin (https://github.com/elmahio/elmah-io-ai-plugin) (Apache 2.0) packages this MCP server together with prebuilt skills (debug-production-error, post-deployment-check, investigate-frequent-errors, check-uptime-status) and a background deployment-watchdog agent that watches for regressions after a deploy.
claude plugin marketplace add elmahio/elmah-io-ai-plugin claude plugin install elmah-io@elmah-io
Links
elmah.io MCP 的远程 MCP 服务器是一个托管的 Model Context Protocol 端点,地址是 https://mcp.elmah.io/mcp,AI 助手无需在本地安装或运行任何东西即可连接。
把端点 https://mcp.elmah.io/mcp 添加到任意支持 MCP 的客户端,比如 Claude Code、Cursor 或 VS Code。本页的接入代码块可以一步配置好每个客户端。
需要。elmah.io MCP 使用 OAuth——第一次连接时,你的 MCP 客户端会打开浏览器窗口登录并授权,之后会复用这个凭证。
elmah.io MCP 暴露的是 Streamable HTTP 端点,这是远程 MCP 服务器普遍使用、主流客户端都支持的传输方式。
Aard connects Macroeconomic and other official data from 170+ publishers including World Bank, IMF, BIS, ECB, Eurostat, and National Statistical Offices. Aard is powered by a large graph of the metadata universe and an
Hosted AdLoop: Google Ads, Reddit Ads, GA4 and Tag Manager in claude.ai, ChatGPT and more, with a preview before every change.
Convert web pages to clean Markdown and retrieval-ready chunks with browser rendering support.
Email for long-term client relationships: ARVX reads each person first and remembers every pair.
Pharmaceutical drug & clinical trial intelligence
评论