covidence-mcp
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@covidence-mcp Screen the next 15 studies in review 12345"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
covidence-mcp
An MCP connector that lets Claude screen studies in Covidence using its own intelligence — no brittle CSS selectors, no hardcoded click paths.
How it works
Instead of a static Playwright script, Claude navigates Covidence directly using Claude in Chrome. It reads the live page, finds the right buttons by understanding what it sees, and casts votes — the same way a human would. When Covidence updates their UI, nothing breaks.
The MCP server itself is intentionally thin: it stores your inclusion/exclusion criteria per review and keeps a session vote log. All actual browser interaction is handled by Claude.
You ──► Claude ──► covidence_screen (MCP)
│
▼
Returns screening prompt
│
▼
Claude navigates Chrome directly
(read_page → reason → find → click)
│
▼
Votes cast in CovidenceRelated MCP server: chrome-bridge
Setup
There are two ways to connect, depending on whether you're using Claude Desktop or the claude.ai web app.
Option A — Claude Desktop (local)
Requirements: Node.js ≥ 18, Claude Desktop app
git clone <this repo>
cd covidence-mcp
npm install
npm run buildAdd to your Claude Desktop config:
macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"covidence": {
"command": "node",
"args": ["/absolute/path/to/covidence-mcp/dist/index.js"]
}
}
}Restart Claude Desktop. Done.
Option B — claude.ai web (remote hosting)
Claude.ai supports remote MCP servers over SSE. You deploy this server somewhere public and give Claude the URL — no desktop app required.
Requirements: A free account on Railway, Render, or any host that can run Node.js
1. Deploy to Railway (easiest)
Or manually:
# Push this folder to a GitHub repo, then:
# 1. Create a new Railway project from that repo
# 2. Railway auto-detects Node.js and runs `npm run build && npm start`
# 3. Set the PORT environment variable (Railway sets this automatically)The server switches to HTTP mode automatically when PORT is set. Your public URL will look like:
https://covidence-mcp-production.up.railway.app2. Connect to claude.ai
Go to claude.ai → Settings → Integrations
Click Add custom connector
Enter your server URL:
https://your-deployment.up.railway.app/sseSave — Claude will confirm the connection
Deploy to Render (alternative)
Create a new Web Service from your GitHub repo
Build command:
npm install && npm run buildStart command:
node dist/index.jsRender sets
PORTautomatically
Deploy to Fly.io (alternative)
fly launch
fly deployThen connect https://your-app.fly.dev/sse in Claude's integrations settings.
Usage
Once connected (either way), the workflow is the same.
First time — tell Claude your login and criteria:
Log in to Covidence with researcher@university.edu, then save these criteria for review 12345:
Include: RCTs and quasi-experimental studies in adults with type 2 diabetes.
Exclude: animal studies, systematic reviews, non-English publications, studies before 2000.Screen a batch:
Screen the next 20 studies in review 12345.Claude calls covidence_screen, opens Covidence in Chrome, reads a batch of abstracts, applies your criteria, and votes on all of them.
Check progress:
How many studies have we screened today?Tools
Tool | What it does |
| Starts a session and returns Chrome navigation steps for login |
| Saves inclusion/exclusion criteria for a review ID |
| Builds a full screening prompt — Claude uses this to drive Chrome |
| Records a vote in the session log |
| Returns all votes cast this session with totals |
| Returns plain-English navigation steps for any specific action |
License
MIT
Available Tools
6 toolscovidence_get_session_logA
Return a summary of all votes cast in the current session.
Inputs: clear.
covidence_loginA
Start a Covidence session. Returns step-by-step instructions for Claude to navigate the login page using Chrome tools. The user's password is never passed through this server.
Inputs: email (required).
covidence_log_voteA
Record a vote in the session log after Claude has cast it in Chrome. Used to track progress across a session.
Inputs: decision (required), study_id (required).
covidence_screenA
Run a screening session. Returns a complete prompt for Claude to use when navigating Covidence in Chrome — covering navigation, reading abstracts, applying criteria, and casting votes. Claude executes this using Chrome tools.
Inputs: review_id (required), batch_size, criteria_override.
covidence_set_criteriaA
Store inclusion/exclusion criteria for a specific review. Claude will apply these automatically during screening sessions.
Inputs: criteria (required), review_id (required).
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 6 tools
Tool boundaries are muddled because covidence_login duplicates the login action in covidence_nav, and covidence_screen returns a full prompt that covers the same reading/voting instructions available via covidence_nav. The remaining tools are distinct, but these overlapping pairs make selection ambiguous.
The consistent covidence_ prefix and snake_case help, but the forms are mixed: login and screen are bare verbs, nav is an abbreviation, while set_criteria, log_vote, and get_session_log follow a verb_noun pattern. This is readable but not a uniform naming convention.
With six tools, the server is well-scoped for its niche purpose of guiding Covidence screening sessions. The count is neither bloated nor too thin, and each tool contributes to the core workflow.
The core screening loop—login, set criteria, screen, log votes, and view the session log—is covered, with nav supplying supporting instructions for Chrome-driven actions. Minor gaps like no explicit criteria retrieval or session-end tool are workable but prevent a perfect score.
Maintenance
Related MCP Connectors
Give AI agents a real cloud Chrome: open sessions, navigate, read pages, click, type and screenshot.
Let AI agents use your real, signed-in Chrome: tabs, clicks, forms, screenshots and video.
Control a real browser from AI agents: navigate, click, type, screenshot, read the page.
Read-only web QA agent: scans a site in a real browser and returns evidence-backed findings
Related MCP Servers
- AlicenseAqualityDmaintenanceAttaches to your existing Chrome browser to allow Claude to see the active tab, interact via accessibility tree, and detect handoff triggers like 2FA and captchas.10MIT
- FlicenseNot gradedqualityCmaintenanceProvides a direct bridge between Claude and an authenticated Chrome session, enabling fetch requests with session cookies and JavaScript execution within live web pages.-
- AlicenseNot gradedqualityBmaintenanceLets Claude Code drive your own Chrome browser with human-like clicks, drags, and typing. It uses your real sessions and logins, and pauses for your confirmation on anything irreversible.8MIT
- AlicenseNot gradedqualityBmaintenanceEnables Claude Code and Claude Desktop to control your own Chrome browser, with Jev deciding each click, keystroke, and scroll.5MIT