Search documentation
Jump to any section of the Crispy docs
Get Started
Connect Crispy to your AI
Crispy gives your AI complete access to LinkedIn: outreach, content, inbox, and your network. Get connected in under two minutes.
Overview
The primary way in is the MCP server. Connect Claude, Cursor, ChatGPT, n8n, or any MCP-compatible client and your AI can take actions on LinkedIn immediately. If you prefer to call Crispy from code, the REST API works from any language with a Bearer token. For one-liners, scripts, and CI pipelines, the CLI is the third surface: the same tools as the MCP server, over stdin/stdout.
New
Before you start
- Create an account on Crispy
- Connect your LinkedIn profile from the dashboard
- Generate an API key (starts with
crispy_) - Copy your API key, it's shown only once
Claude.ai and ChatGPTuse OAuth and don't need an API key, just paste the endpoint URL and log in. All other platforms use an API key.
Connect your tool
Pick your platform. Each one connects to the same MCP server, so your tools and limits are identical everywhere.
Connect Crispy directly from Claude.ai using the built-in connector system. Claude handles authentication via OAuth automatically.
- Go to Claude.ai → Settings → Connectors
- Click Add Custom Connector
- Enter the MCP endpoint URL:
https://crispy.sh/api/mcp- Claude redirects you to Crispy to log in and pick a LinkedIn account
- Click Approve on the consent screen
- Done — Crispy tools are available in all your conversations
Connect Crispy as a custom MCP server via Developer mode. Available on ChatGPT Pro, Plus, Business, Enterprise, and Edu plans on the web.
- Go to ChatGPT → Settings → Apps & Connectors
- Open Advanced settings and turn on Developer mode
- Back in Apps & Connectors, click Create appand give it a name (e.g. "Crispy")
- Set the MCP Server URL:
https://crispy.sh/api/mcp- Set Authentication to OAuth
- Approve the consent screen after logging in to Crispy
- In any chat, open the + menu → Developer mode and select Crispy
Two ways to connect. Pick whichever you prefer.
Option A: OAuth (recommended)
Claude Code discovers the OAuth flow automatically and opens your browser to approve. No API key to manage.
claude mcp add --transport http crispy https://crispy.sh/api/mcpOption B: API key
Faster if you already have a key. No browser needed.
claude mcp add crispy -- \
--url "https://crispy.sh/api/mcp" \
--header "Authorization: Bearer YOUR_API_KEY"OpenAI's Codex CLI has first-class support for OAuth-protected MCP servers. The OAuth flow runs automatically the first time you connect.
codex mcp add crispy --url https://crispy.sh/api/mcpCodex prints a URL — open it, sign in to crispy.sh, click Approve. Tokens are stored in ~/.codex and refreshed automatically. Prefer a key? Use --bearer-token-env-var CRISPY_API_KEY.
Open the command palette and run MCP: Open User Configuration, then add this to .vscode/mcp.json:
{
"servers": {
"crispy": {
"url": "https://crispy.sh/api/mcp",
"type": "http",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}Requires GitHub Copilot with agent mode enabled.
Go to Cursor Settings → MCP → Add Server, and paste this config:
{
"mcpServers": {
"crispy": {
"url": "https://crispy.sh/api/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}Cursor uses an API key in the headers block rather than the auto-OAuth flow. Generate one in your dashboard.
Add this to your Windsurf MCP config file (~/.codeium/windsurf/mcp_config.json):
{
"mcpServers": {
"crispy": {
"serverUrl": "https://crispy.sh/api/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}Windsurf uses serverUrl instead of url for remote HTTP servers.
Works with IntelliJ IDEA, WebStorm, PyCharm, and other JetBrains IDEs with AI Assistant. Go to Settings → AI Assistant → MCP Servers and add:
{
"mcpServers": {
"crispy": {
"url": "https://crispy.sh/api/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}Use the MCP Client node to connect any n8n workflow to Crispy.
1. Add an "MCP Client" node to your workflow
2. Set the Server URL to: https://crispy.sh/api/mcp
3. Add a header: Authorization = Bearer YOUR_API_KEY
4. The node will auto-discover all available toolsUse the built-in MCP Client module to connect Make scenarios. Available on all paid plans.
- Add an MCP Client module to your scenario
- Enter the server URL:
https://crispy.sh/api/mcp- Set authentication to Token and paste your API key
- Make auto-discovers all tools with input schemas
Use the MCP Client by Zapierapp to connect any Zap to Crispy's tools.
- Add an MCP Client by Zapier step to your Zap
- Choose Run Tool as the action
- Enter the server URL:
https://crispy.sh/api/mcp- Add your API key as a Bearer token in the authorization header
- Select the Crispy tool you want and map your inputs
The MCP endpoint speaks JSON-RPC over HTTP. Call it directly from any language. See the protocol reference for the full method list.
curl -X POST https://crispy.sh/api/mcp \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'Your first call
Once connected, verify everything works by listing the available tools. From an AI client, just ask it to "list Crispy tools." From the command line:
curl -X POST https://crispy.sh/api/mcp \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"jsonrpc": "2.0",
"id": 1,
"method": "tools/list"
}'A successful response returns the full tool catalog with input schemas. Ready to do something real? Try search_people or run the inbox_zero prompt. See the tool reference for the full list.
Next steps
Concepts
→Permission scopes, multi-account, and the safety limits that protect your LinkedIn account.
API Reference
→The JSON-RPC protocol, every tool and prompt, resources, and error codes.
Campaigns
→Run automated multi-step sequences or track manual outreach for analytics.
Webhooks
→Get real-time signed events when things happen on your LinkedIn account.