Scrapely MCP Server
An MCP (Model Context Protocol) server that wraps the Scrapely API, letting AI assistants like Claude directly manage your Twitter/X DM outreach — campaigns, lead scraping, conversations, CRM, and scheduled tweets — all through natural language.
GitHub: github.com/deepblue6/scrapely-mcp
Install
npm install -g scrapely-mcp
Or run it directly without installing:
npx scrapely-mcp
Setup
You need a Scrapely API key. Generate one from Settings in your Scrapely dashboard.
Claude Code
claude mcp add scrapely -e SCRAPELY_API_KEY=sk_live_your_key_here -- npx scrapely-mcp
Claude Desktop
Edit ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):
{
"mcpServers": {
"scrapely": {
"command": "npx",
"args": ["scrapely-mcp"],
"env": {
"SCRAPELY_API_KEY": "sk_live_your_key_here"
}
}
}
}
Other MCP clients
Any MCP-compatible client works. Just run npx scrapely-mcp with SCRAPELY_API_KEY set in the environment.
Available Tools
| Tool | Description |
|---|---|
list_accounts |
List all connected Twitter/X accounts |
add_account |
Add a new account (cookies + proxy) |
remove_account |
Remove an account by ID or handle |
update_cookies |
Refresh cookies for an existing account |
update_proxy |
Update proxy settings for an account |
toggle_pause |
Pause/unpause DM sending |
toggle_followups_pause |
Pause/unpause follow-up messages |
get_account_tags |
Get tags for accounts |
set_account_tags |
Replace all tags for an account |
update_account_tags |
Add or remove individual tags |
create_scraping_source |
Scrape followers/following with keyword and count filters |
check_scraping_status |
Check progress of scraping jobs |
launch_campaign |
Launch a DM campaign with A/B testing, auto-follow/like/comment |
list_campaigns |
List campaigns with stats |
get_campaign_analytics |
Detailed analytics: reply rates, sentiment, variant stats. Supports date range filtering. |
fetch_conversations |
Fetch conversations with optional message history |
send_dm |
Send a DM to an existing conversation |
get_crm_data |
Get conversations organized by CRM tags |
update_crm |
Update notes, deal value, and tags |
enrich_twitter |
Find a person's Twitter profile from name + company |
enrich_email |
Find a person's email from their Twitter handle |
get_webhook_info |
Get webhook event types and payload structure docs |
schedule_tweet |
Schedule a tweet for later |
list_scheduled_tweets |
List scheduled tweets with filters |
cancel_scheduled_tweet |
Cancel a pending scheduled tweet |
Example Prompts
Once connected to Claude, you can say things like:
- "Show me all my connected accounts"
- "Scrape followers of @naval and @paulg, filter for founders with 1k+ followers"
- "Launch a campaign called Q1 Outreach targeting my Tech Founders source"
- "What are my campaign analytics? Which variant is performing best?"
- "Show me campaign stats from January 1st to January 31st"
- "Show me all interested replies in the CRM"
- "Send a followup to conversation 123456-789012"
- "Schedule a tweet from my main account for tomorrow at 9am"
- "Find the Twitter profile for John Doe at Acme Corp"
- "Find the email for @elonmusk"
- "Pause DM sending on my backup account"
Testing with MCP Inspector
SCRAPELY_API_KEY=sk_live_your_key npx @modelcontextprotocol/inspector npx scrapely-mcp
This opens a browser GUI where you can test every tool before connecting it to Claude.
Requirements
- Node.js 18+
- A Scrapely account with an API key
License
MIT