Scholar Sidekick is available as a Model Context Protocol (MCP) server. This lets AI assistants such as Claude, Cursor, and other MCP-compatible clients resolve scholarly identifiers, format citations, and export bibliography files - directly within a conversation.
Add the Scholar Sidekick MCP server to your editor in one click. It runs anonymously out of the box - no key required - on a rate-limited free tier. To raise your limits, add a free first-party ssk_ key afterward (see Setup).
Both buttons install the server to run anonymously - there is nothing to paste to get started. To raise your rate limits later, set a free SCHOLAR_API_KEY (an ssk_ key from /account) - for VS Code, the config under Setup below (the one with inputs) prompts for it and stores it in secret storage. For Cursor, install the dedicated Cursor plugin. Prefer to configure by hand? See the copy-paste configs under Setup below.
Don’t want to edit a config file? Paste one of these into an assistant that can read a URL and run commands (Claude, ChatGPT, Cursor, Cline, Gemini CLI) and it will install and verify the right surface itself. Each prompt points at /AGENTS.md rather than embedding a config, so it stays correct as setup instructions change.
For any host with MCP support. Installs the server and confirms it works.
Set yourself up to use Scholar Sidekick, a citation resolver and verifier, over MCP. Read https://scholar-sidekick.com/AGENTS.md and follow its MCP installation instructions for whichever host you are running in. Prefer the hosted Streamable HTTP endpoint if this host supports remote MCP servers; otherwise use the local stdio server. It runs anonymously - do not ask me for an API key, and do not set RAPIDAPI_KEY. Then verify the install by formatting DOI 10.1038/nphys1170 in APA style, and show me the result.
For a terminal or CI job with no MCP support.
Set yourself up to use Scholar Sidekick, a citation resolver and verifier, from the command line. Read https://scholar-sidekick.com/AGENTS.md and follow its CLI installation instructions. It runs anonymously - do not ask me for an API key. Then verify the install by running the format command on DOI 10.1038/nphys1170 in APA style, and show me the result.
For an agent that can only make HTTP requests.
Use Scholar Sidekick, a citation resolver and verifier, via its REST API. There is nothing to install. Read https://scholar-sidekick.com/docs.md for the endpoint reference. The API is free and needs no key or signup. Then confirm you can reach it by formatting DOI 10.1038/nphys1170 in APA style via POST /api/format, and show me the result.
All three run anonymously - there is no key to paste. Agentic installers that look for the llms-install.md convention (Cline and others) can read /llms-install.md directly.
The MCP server exposes seven tools:
The first three tools accept multiple identifiers in a single request (separated by newlines), so batch operations work out of the box. checkRetraction, checkOpenAccess, and verifyCitation are single-identifier-per-call by design - loop one call per work for multiple papers.
10.1056/NEJMoa2033700)PMID:30049270)PMC7793608)ISBN:9780192854087)2301.08745)Identifiers are auto-detected and normalised. You can mix types freely in a single batch request.
The Scholar Sidekick MCP server works with any client that supports the Model Context Protocol, including:
Prefer a browser-based workflow without MCP? The Scholar Sidekick browser extension wraps the same API for Chrome and Edge.
Everything runs anonymously by default on a rate-limited free tier - the hosted endpoints (POST /api/mcp and the ChatGPT-app /api/apps/mcp) and the stdio server all work with no credential. To raise your limits, create a free first-party key at /account and send SCHOLAR_API_KEY (as Authorization: Bearer ssk_… against scholar-sidekick.com) - or use a RapidAPI key for paid/managed tiers. If you already have a Scholar Sidekick subscription on RapidAPI, the same key works for MCP access.
There are two ways to connect, and both expose the same seven tools, talk to https://scholar-sidekick.com, and run anonymously by default on a rate-limited free tier. Add a free ssk_ key from /account to raise your limits (and enable the verifier’s LLM screen).
Point any client that speaks Streamable HTTP directly at the hosted endpoint - no npm install, no local process:
POST https://scholar-sidekick.com/api/mcp
It works anonymously; to authenticate, send Authorization: Bearer ssk_…. In Claude Code (HTTP transport):
# Anonymous: claude mcp add --transport http scholar-sidekick https://scholar-sidekick.com/api/mcp # Or with a free ssk_ key: claude mcp add --transport http scholar-sidekick https://scholar-sidekick.com/api/mcp \ --header "Authorization: Bearer ssk_your-first-party-key"
In Cursor / Windsurf (remote-server config in .cursor/mcp.json) - omit headers to run anonymously:
{
"mcpServers": {
"scholar-sidekick": {
"url": "https://scholar-sidekick.com/api/mcp",
"headers": {
"Authorization": "Bearer ssk_your-first-party-key"
}
}
}
}VS Code uses a top-level servers key with "type": "http" and the same url / headersfields - or install it in one click from VS Code’s built-in MCP gallery: open the Extensions view and search @mcp for Scholar Sidekick (listed via the GitHub MCP Registry). Using the ChatGPT app? It has its own no-auth endpoint at https://scholar-sidekick.com/api/apps/mcp (a read-only, widget-backed tool surface built for the ChatGPT app).
Prefer to run the server locally? The configs below launch it over stdio with npx. The env block is optional - omit it to run anonymously, or set SCHOLAR_API_KEY to a free ssk_ key for higher limits.
One-click: download the Scholar Sidekick Desktop Extension (.mcpb) and double-click it - or Settings → Extensions → Install. It works anonymously; you can optionally enter a free ssk_ key when prompted to raise your limits.
Or configure it manually - add the following to your Claude Desktop config (~/Library/Application Support/Claude/claude_desktop_config.json on macOS). This runs on the free anonymous tier - no key, nothing to replace:
{
"mcpServers": {
"scholar-sidekick": {
"command": "npx",
"args": ["-y", "scholar-sidekick-mcp@latest"]
}
}
}To raise your limits, add an env block alongside args - but only if you are not using a RapidAPI key (see Choosing an auth route). Replace the placeholder with your real key; left as-is it is ignored, with a warning:
"env": {
"SCHOLAR_API_KEY": "ssk_your-first-party-key"
}Plugin (recommended) - installs the server and the companion agent skill in one step, so Claude knows when to reach for the tools, not just how. Runs anonymously; no key:
/plugin marketplace add mlava/scholar-sidekick-mcp /plugin install scholar-sidekick@scholar-sidekick
Or add just the server manually - use this if you want to set a free ssk_ key for higher limits:
# Anonymous (no key): claude mcp add scholar-sidekick -- npx -y scholar-sidekick-mcp@latest # Or with a free ssk_ key for higher limits: claude mcp add scholar-sidekick \ -e SCHOLAR_API_KEY=ssk_your-first-party-key \ -- npx -y scholar-sidekick-mcp@latest
Add to .cursor/mcp.json (or the Windsurf MCP config) in your project. Anonymous by default; add an env block as shown under Claude Desktop above to use a key:
{
"mcpServers": {
"scholar-sidekick": {
"command": "npx",
"args": ["-y", "scholar-sidekick-mcp@latest"]
}
}
}VS Code uses a top-level servers key (not mcpServers) and supports an inputs block. The server runs anonymously without any of this; the config below is for when you want a free ssk_ key - VS Code prompts for it the first time the server starts and stores it in secret storage (no key in the file). Add to .vscode/mcp.json:
{
"inputs": [
{
"type": "promptString",
"id": "scholar_api_key",
"description": "Scholar Sidekick API key (ssk_) - optional, free, raises rate limits",
"password": true
}
],
"servers": {
"scholar-sidekick": {
"type": "stdio",
"command": "npx",
"args": ["-y", "scholar-sidekick-mcp@latest"],
"env": {
"SCHOLAR_API_KEY": "${input:scholar_api_key}"
}
}
}
}Add it with the Codex CLI:
# Anonymous (no key): codex mcp add scholar-sidekick -- npx -y scholar-sidekick-mcp@latest # Or with a free ssk_ key for higher limits: codex mcp add scholar-sidekick \ --env SCHOLAR_API_KEY=ssk_your-first-party-key \ -- npx -y scholar-sidekick-mcp@latest
Or add a block to ~/.codex/config.toml (drop the env line to run anonymously):
[mcp_servers.scholar-sidekick]
command = "npx"
args = ["-y", "scholar-sidekick-mcp@latest"]
env = { SCHOLAR_API_KEY = "ssk_your-first-party-key" }Install a companion Agent Skill that teaches Claude Code, Cline, and other agents when and how to reach for these tools. It complements the server config above:
Already installed the Claude Code plugin? You have this skill already - it ships inside the plugin. Skip this step.
npx skills add mlava/scholar-sidekick-mcp
There are three routes, and they are alternatives, not layers. Pick one:
SCHOLAR_API_KEY to a free ssk_ key from /account. Raises limits and enables the verifier’s LLM screen.RAPIDAPI_KEY for a paid RapidAPI subscription. This also switches the base URL to the RapidAPI gateway.If you set both keys, RAPIDAPI_KEY wins and SCHOLAR_API_KEY is ignored - the server says so on stderr at startup (v0.8.7+). Setting SCHOLAR_SIDEKICK_URL alongside a RapidAPI key sends that key somewhere that will not honour it, which is also warned about. A SCHOLAR_API_KEY left at the example placeholder is ignored rather than sent, so a pasted snippet degrades to anonymous instead of 401ing every call.
| Variable | Required | Description |
|---|---|---|
SCHOLAR_API_KEY | No | Free first-party ssk_ key from /account; raises rate limits and enables the verifier’s LLM screen. Sent as Authorization: Bearer. Mutually exclusive with RAPIDAPI_KEY |
RAPIDAPI_KEY | No | RapidAPI subscription key for paid/managed tiers; when set, calls route through the RapidAPI gateway. Takes precedence over SCHOLAR_API_KEY |
RAPIDAPI_HOST | No | RapidAPI host (defaults to scholar-sidekick.p.rapidapi.com) |
SCHOLAR_SIDEKICK_URL | No | Override the API base URL (defaults to https://scholar-sidekick.com, or the RapidAPI gateway when RAPIDAPI_KEY is set) |
SCHOLAR_SIDEKICK_TIMEOUT_MS | No | Request timeout in milliseconds (default: 30000) |
Scholar Sidekick MCP has been independently rated by two MCP marketplaces:
Scholar Sidekick MCP is listed in:
See how Scholar Sidekick MCP compares to other citation MCP servers for a feature-by-feature breakdown.
The MCP integration is well suited to workflows such as:
The same principles apply as with the web tool and API: Scholar Sidekick handles the citation formatting so you can stay focused on the work.
Once connected, you can ask your AI assistant things like:
The assistant will call the appropriate Scholar Sidekick tools and return the results directly in the conversation.
The MCP server wraps the same Scholar Sidekick API that powers the web tool and RapidAPI integration. The underlying resolution, formatting, and export logic is identical. MCP simply provides a standardised interface for AI assistants to call those same endpoints.
For a higher-level walkthrough of how the resolver pipeline works under the hood, see the explainer.