Altmetric MCP Server
Provides access to Altmetric APIs for monitoring research impact and attention across news, policy documents, social media, and other online platforms.
README
Altmetric MCP Server
Model Context Protocol (MCP) server that enables AI agents to access Altmetric APIs for tracking the attention and reach of research outputs across news outlets, policy documents, social media, and other online platforms.
Altmetric monitors where research is being discussed beyond traditional academic citations - from mainstream media coverage to policy citations, patent references, and social media engagement - providing a comprehensive view of real-world research impact.
Prerequisites
- Node.js >= 20.6.0 - Download from nodejs.org (LTS recommended)
- Altmetric API credentials (at least one):
- Details Page API key - Free tier or commercial access
- Explorer API key + secret - Institutional access
Don't have keys yet? Request API access
Quick Install (Claude Desktop on macOS)
Run the guided installer in Terminal - it checks Node.js, prompts for your API keys, and configures Claude Desktop automatically:
bash <(curl -fsSL https://raw.githubusercontent.com/altmetric/altmetric-mcp/main/install.sh)
Or if you've cloned the repo:
bash install.sh
Prefer to set things up manually? See Claude Desktop below.
Installation
Configure your MCP client to run the Altmetric MCP server using npx. Include only the API credentials you have access to.
{
"type": "stdio",
"command": "npx",
"args": ["-y", "altmetric-mcp"],
"env": {
"ALTMETRIC_DETAILS_API_KEY": "your_details_api_key_here",
"ALTMETRIC_EXPLORER_API_KEY": "your_explorer_api_key_here",
"ALTMETRIC_EXPLORER_API_SECRET": "your_explorer_api_secret_here"
}
}
Below are specific instructions for popular AI tools and editors.
Claude Desktop
-
Open the configuration file at:
- macOS:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
%APPDATA%\Claude\claude_desktop_config.json
On macOS you can open it from Terminal:
mkdir -p ~/Library/Application\ Support/Claude && open -a TextEdit ~/Library/Application\ Support/Claude/claude_desktop_config.json - macOS:
-
Add the Altmetric MCP server. If the file is empty, paste this (replacing the placeholder keys with your own, and removing any you don't have):
{ "mcpServers": { "Altmetric": { "type": "stdio", "command": "npx", "args": ["-y", "altmetric-mcp"], "env": { "ALTMETRIC_DETAILS_API_KEY": "your_details_api_key_here", "ALTMETRIC_EXPLORER_API_KEY": "your_explorer_api_key_here", "ALTMETRIC_EXPLORER_API_SECRET": "your_explorer_api_secret_here" } } } }If the file already has content, add
"Altmetric": { ... }inside the existing"mcpServers"block, separated by a comma from the other entries. -
Save the file and restart Claude Desktop (Cmd+Q then reopen).
-
Verify by asking Claude: "Use the Altmetric tools to look up the attention score for DOI 10.1038/nature12373"
<details> <summary><strong>Claude Code</strong></summary>
Install directly from the command line:
claude mcp add --transport stdio altmetric-mcp \
--env ALTMETRIC_DETAILS_API_KEY=your_details_api_key_here \
--env ALTMETRIC_EXPLORER_API_KEY=your_explorer_api_key_here \
--env ALTMETRIC_EXPLORER_API_SECRET=your_explorer_api_secret_here \
-- npx -y altmetric-mcp
</details>
<details> <summary><strong>VS Code (GitHub Copilot)</strong></summary>
Add to your project .vscode/mcp.json:
{
"servers": {
"Altmetric": {
"type": "stdio",
"command": "npx",
"args": ["-y", "altmetric-mcp"],
"env": {
"ALTMETRIC_DETAILS_API_KEY": "your_details_api_key_here",
"ALTMETRIC_EXPLORER_API_KEY": "your_explorer_api_key_here",
"ALTMETRIC_EXPLORER_API_SECRET": "your_explorer_api_secret_here"
}
}
}
}
Reload VS Code to apply the changes. More information.
</details>
<details> <summary><strong>Cursor</strong></summary>
Add to ~/.cursor/mcp.json:
{
"mcpServers": {
"Altmetric": {
"type": "stdio",
"command": "npx",
"args": ["-y", "altmetric-mcp"],
"env": {
"ALTMETRIC_DETAILS_API_KEY": "your_details_api_key_here",
"ALTMETRIC_EXPLORER_API_KEY": "your_explorer_api_key_here",
"ALTMETRIC_EXPLORER_API_SECRET": "your_explorer_api_secret_here"
}
}
}
}
</details>
<details> <summary><strong>Other MCP clients</strong></summary>
Any MCP-compatible client that supports stdio transport can use this server. Use the generic configuration at the top of this section, adapting it to your client's config format. The command is always npx with args ["-y", "altmetric-mcp"] plus the environment variables for your API keys.
</details>
Deploying safely
This server runs as a child process of the MCP host (Claude Desktop, Claude Code, etc.). A few things are worth knowing before you wire it into a sensitive workflow.
What the server does
- Read-only proxy to two Altmetric HTTP APIs over outbound HTTPS. The one exception is an idempotent
POSTto the Explorer identifier_lists endpoint (create-or-find), used internally to scope an Explorer query to a supplied set of identifiers; it creates no user-visible state and is not destructive. - No inbound network surface; no destructive operations.
- Treats upstream text as untrusted: scans for prompt-injection markers, redacts suspicious matches in the LLM-facing summary, and surfaces raw values only via
structuredContent.
What you should do
- Set API keys via your MCP host's
envblock, not via a committed.envfile. Keys appear in URL query strings; treat them as bearer-equivalent credentials. - Run with reduced privileges if your host allows it. The server only needs outbound HTTPS to
api.altmetric.comandwww.altmetric.com. If your host or container runtime supports it, deny filesystem writes outside$TMPDIRand deny other network egress. - Consider an egress allowlist (forward proxy / DLP). If you're using the server inside an environment that processes sensitive data, route outbound traffic through a proxy that only allows the two Altmetric hosts. The server doesn't need to talk to anyone else.
- Respect your data-classification zone. This server forwards tool arguments verbatim to a third party. If your prompt contains restricted or regulated data, it leaves your boundary.
What's enforced for you
- Outbound URLs are asserted to use
https:. - Upstream responses are capped at 20 MB and 60 s.
- Inbound tool arguments are capped at 8 MB total / 64 KB per string.
- Filter values are validated client-side (date format, length, pagination ranges) before any upstream call.
- Upstream error bodies are logged to stderr by SHA-256 prefix only, not verbatim.
For vulnerability reports and supported versions see SECURITY.md.
Troubleshooting
| Problem | Solution |
|---|---|
command not found: node |
Node.js is not installed. Download it here (version 20.6.0 or later). |
| Claude Desktop won't start after editing config | The JSON file has a syntax error. Check for missing commas, unmatched brackets, or trailing commas. Paste it into jsonlint.com to validate. |
| "MCP server failed to start" | Run npx -y altmetric-mcp in Terminal to see the actual error. Usually a missing/invalid API key or Node.js version too old. |
| Tools appear but return 403 errors | You're using a free-tier key with a commercial-tier tool (get_citation_details). Use get_citation_counts or search_citations instead. |
| First query is slow | Normal. npx downloads the package on first run. Subsequent uses are faster. |
| Explorer tools fail | Explorer tools need both ALTMETRIC_EXPLORER_API_KEY and ALTMETRIC_EXPLORER_API_SECRET. Make sure both are set. |
API Tiers
Details Page API
- Free Tier: Access to
get_citation_countsandsearch_citations - Commercial Tier: Access to all Details Page tools including
get_citation_details
If you attempt to use get_citation_details with a free API key, you'll receive a 403 error.
Explorer API
- Institutional Access Only: All Explorer API tools require institutional credentials
- Provides access to research outputs and attention data across the entire Altmetric database
- Organizations with data integrations can also access their own institutional research outputs in isolation
- Includes advanced filtering by author, department, journal, and custom organizational metadata
Tools
This server provides eleven tools across two APIs:
| Tool | API | Tier | Description |
|---|---|---|---|
get_citation_counts |
Details Page | Free | Attention metrics by identifier (DOI, PubMed ID, etc.) |
get_citation_details |
Details Page | Commercial | Full mention text, author details (incl. Dimensions Researcher IDs), metadata |
search_citations |
Details Page | Free | Search attention data across all outputs by timeframe |
get_batch_attention_data |
Details Page | Commercial | Attention metrics for many DOIs at once, ranked |
translate_identifiers |
Details Page | Commercial | Translate identifiers (DOI, PMID, etc.) to Altmetric IDs |
explore_research_outputs |
Explorer | Institutional | Search and filter research outputs |
explore_attention_summary |
Explorer | Institutional | Aggregated attention metrics by source and date |
explore_mentions |
Explorer | Institutional | Individual mention details with filtering |
explore_demographics |
Explorer | Institutional | Audience geographic and demographic data |
explore_mention_sources |
Explorer | Institutional | Source/outlet analysis for mentions |
explore_journals |
Explorer | Institutional | Journal metrics, rankings, and search |
All Explorer tools additionally accept researcher_id and grant_id filters (Dimensions IDs), and an identifiers parameter that scopes a query to a raw list of scholarly identifiers - the server builds the corresponding identifier list for you. Explorer responses also include sentiment data (sentiment-analysis-totals on research outputs, sentiment-analysis on X/Bluesky mentions).
For detailed parameters and examples, see TOOLS.md.
API Documentation
- Details Page API: https://docs.altmetric.com/details-page-api/
- Explorer API: https://docs.altmetric.com/explorer-api/
License
MIT
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。