Peliqan MCP Server

What is the Peliqan MCP?

The Peliqan MCP (Model Context Protocol) server lets AI assistants like Claude Code and Claude Desktop talk directly to your Peliqan account. Once connected, the AI can explore your data, debug pipelines, write and run data apps, manage query tables, and call API endpoints — all without leaving your chat. Think of it as giving Claude a direct line into Peliqan.

Prerequisites

• A Peliqan account (e.g. EU region: app.eu.peliqan.io)
• A Peliqan API token — generate one from your Peliqan account settings
• Claude Code or Claude Desktop installed

Setup

Claude Code (terminal)

Add the following to your ~/.claude.json file. If the file doesn't exist yet, create it.

{
  "mcpServers": {
    "peliqan": {
      "type": "http",
      "url": "https://mcp.eu.peliqan.io/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_PELIQAN_TOKEN"
      }
    }
  }
}

Replace YOUR_PELIQAN_TOKEN with your personal Peliqan API token.

After saving, restart Claude Code. You should see peliqan listed when you run /mcp in the chat.

Claude Desktop

Open Claude Desktop settings → Developer → Edit Config and add the below block under mcpServers:

"mcpServers": {
    "peliqan": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "https://mcp.eu.peliqan.io/mcp",
        "--header",
        "Authorization: Bearer TOKEN"
      ]
    }
  }

Make sure npx and mcp-remote are installed on your computer. For Mac:

• brew install node
• npx mcp-remote

Other MCP-compatible clients

Any MCP-compatible AI client that supports HTTP transport can connect using the same URL and token. Check your client's documentation for where to add MCP server config.

Verifying the connection

Once connected, try these prompts to confirm everything is working:

"List my Peliqan connections" "Show me all databases in my Peliqan account” "List my data apps"

If you see results from your Peliqan account, you're all set.

What the MCP Server can do

Debug connections & pipelines

• See all your data source connections and their sync status
• View recent pipeline run history
• Read raw error logs when a sync fails

Explore your data

• Browse all databases, schemas, and tables
• Inspect column names and types
• Preview rows of data

Manage query tables

• View run history and error logs for a failing query table
• Check upstream/downstream dependencies before making changes
• Create, update, and delete SQL query tables
• Test fixes in an isolated schema before applying to production

Build & run data apps

• Write Python scripts and run them as background jobs (shell mode)
• Create interactive Streamlit web UIs and publish them with a shareable URL
• Build HTTP API handlers and expose them as REST endpoints
• Schedule scripts to run automatically

Manage API endpoints

• Create, update, and delete HTTP endpoints backed by your Python scripts
• View recent call logs for any endpoint

Sub-account management (partner accounts only)

• List all customer sub-accounts
• Run any tool against a specific sub-account by passing sub_account_id

Resources the AI can read

These are automatically available to the AI — no extra setup needed.

Example prompts to get started

Tips

• EU vs other regions — the MCP URL above is for the EU region. If your account is on a different region or a self-hosted instance, update the hostname accordingly.
• The AI reads docs automatically — before writing any data app script, tell the AI to read peliqan://docs/data-app-pq first. This ensures it uses the correct pq functions and script structure.
• Deletions are permanent — deleting a schema, table, or data app via the MCP cannot be undone. The AI will always confirm before doing this, but double-check before approving.
• API endpoint URLs follow the pattern: https://api.eu.peliqan.io/{accountId}/{route}

Getting your API token

1. Log in to Peliqan at https://app.eu.peliqan.io
2. Go to Admin > Account Settings
3. Go to the API token section
4. Copy your API token
5. Paste it into your MCP config as shown in the Setup section above

Need help?

Reach out to Peliqan support: support@peliqan.io. You can also ask Claude directly — once the MCP is connected, it has full context on how Peliqan works.