Fivetran Mcp

io.github.YimingYANPublished by GitHub account YimingYANv0.3.1Updated Oct 24, 2026

MCP server for Fivetran API - manage syncs, check status, and control data pipelines

VerifiedSTDIODesktop onlyCloud & InfrastructureData & Analytics

Overview

AI-generated overview

Local MCP server for the Fivetran API that lets an assistant list connections, check sync status, and trigger, pause, or resync data pipelines.

What you can do
  • List connections and destination groups, with optional filtering by group
  • Check connection status, tasks, warnings, and run diagnostic connection tests
  • Trigger syncs, full historical resyncs, or resyncs of specific tables
  • Pause and resume connection scheduling
  • Inspect schemas, tables, and columns, and reload schema configuration from the source
When to use it
Useful when you work with Fivetran-managed data pipelines and want to inspect or operate them conversationally — checking whether a connector is healthy, starting a sync, or reviewing which tables and columns are enabled — without opening the Fivetran dashboard.
Requirements
Runs locally as a Python package (fivetran-mcp, launched via uvx) over stdio. Requires a Fivetran API key and secret, supplied through the FIVETRAN_API_KEY and FIVETRAN_API_SECRET environment variables (alternative names FIVETRAN_SYNC_API_KEY and FIVETRAN_SYNC_API_SECRET are also supported). Network access to the Fivetran API is needed.
Before you install
The server holds Fivetran API credentials that grant control over your data pipelines; store them carefully and note the secret is shown only once. Several tools change state: trigger_sync, trigger_resync, resync_tables, pause_connection, resume_connection, and reload_schema can start costly or disruptive jobs. Sync history, logs, and data types are not exposed by the API, so the assistant cannot report on past runs.

Installation

In SourceWeft

  1. Open Fivetran Mcp in the dashboard and add it to a workspace.
  2. Enable the server for the chats that should use its tools.

Desktop only via STDIO. STDIO servers start a local process, so they need the SourceWeft desktop host.

Other MCP clients

Follow the launch instructions in the repository.

README

Fivetran MCP Server

[PyPI version] [License: MIT] [Python 3.10+] [Built by Claude Code]

Note: This repository is built and maintained entirely by Claude Code, Anthropic's AI coding assistant.

An MCP (Model Context Protocol) server for controlling Fivetran data pipelines. Enables AI assistants like Claude to manage syncs, monitor connection status, and control data pipelines through natural language.

Features

  • List connections - View all Fivetran connections with status
  • Check sync status - Get detailed status for any connection including tasks and warnings
  • Trigger syncs - Start syncs on demand
  • Historical resync - Trigger full data resync or resync specific tables
  • Pause/Resume - Control connection scheduling
  • List groups - View all destination groups
  • Test connection - Diagnose connectivity and configuration issues
  • Schema visibility - View schemas, tables, and columns for any connection
  • Reload schema - Refresh schema configuration from the source

Quick Start

Step 1: Get Fivetran API Credentials

  1. Log in to Fivetran Dashboard
  2. Click your username (top right corner)
  3. Click API Key
  4. Click Generate API key
  5. Copy both the API Key and API Secret (secret shown only once!)

See Fivetran API Getting Started for more details.

Step 2: Store Credentials

Add to your ~/.env.local (or equivalent):

bash
# Fivetran APIexport FIVETRAN_API_KEY="your-api-key"export FIVETRAN_API_SECRET="your-api-secret"

Then reload:

bash
source ~/.env.local

Step 3: Verify Credentials

Test that your credentials work:

bash
curl -s -X GET "https://api.fivetran.com/v1/account/info" \  -H "Accept: application/json" \  -H "Authorization: Basic $(echo -n "$FIVETRAN_API_KEY:$FIVETRAN_API_SECRET" | base64)"

Expected response:

json
{"code":"Success","data":{"account_id":"...","account_name":"..."}}

Step 4: Configure Claude Code

Add to your ~/.claude.json:

json
{  "mcpServers": {    "fivetran": {      "type": "stdio",      "command": "uvx",      "args": ["fivetran-mcp@latest"],      "env": {        "FIVETRAN_API_KEY": "${FIVETRAN_API_KEY}",        "FIVETRAN_API_SECRET": "${FIVETRAN_API_SECRET}"      }    }  }}

Step 5: Restart Claude Code

Restart Claude Code to load the new MCP server. You should now have access to Fivetran tools.

Available Tools

Connection Management

ToolDescription
list_connectionsList all connections, optionally filtered by group
get_connection_statusGet detailed status including tasks, warnings, and sync details
trigger_syncStart a sync for a connection (optional force flag)
trigger_resyncTrigger full historical resync
resync_tablesResync specific tables only (e.g., ["schema.table_name"])
pause_connectionPause a connection
resume_connectionResume a paused connection
list_groupsList all groups/destinations
test_connectionRun diagnostic tests to identify connectivity/configuration issues

Schema & Table Visibility

ToolDescription
get_connection_schemaGet schema info with optional table filter (includes columns when filtering)
get_schemaGet complete raw schema configuration for a connection
list_tablesList all tables with enabled status and sync mode
get_table_columnsGet column details for a specific table
reload_schemaRefresh schema configuration from the source

Limitations

The following capabilities are not available via Fivetran's REST API:

CapabilityStatusNotes
Column data typesNot availableQuery your destination database directly for type info
Sync history/logsNot availableFivetran uses Log Service which streams to external destinations
Detailed error logs for past syncsNot availableOnly current tasks/warnings are accessible
Row-level sync statusNot availableNot exposed via API
Data quality metricsNot availableNot exposed via API

For sync logs and historical data, consider setting up Fivetran's Log Service to stream logs to your data warehouse.

Environment Variables

The server supports two naming conventions:

PreferredAlternative
FIVETRAN_API_KEYFIVETRAN_SYNC_API_KEY
FIVETRAN_API_SECRETFIVETRAN_SYNC_API_SECRET

Development

bash
# Clone the repositorygit clone https://github.com/YimingYAN/fivetran-mcp.gitcd fivetran-mcp
# Install dependenciesuv sync
# Run the server locallyuv run fivetran-mcp

Cloud Deployment

Cloudflare Workers (Free)

Deploy to Cloudflare Workers for a hosted MCP endpoint:

bash
cd cloudflareuv tool install workers-pynpx wrangler loginnpx wrangler secret put FIVETRAN_API_KEYnpx wrangler secret put FIVETRAN_API_SECRETuv run pywrangler deploy

See cloudflare/README.md for detailed instructions.

License

MIT

Source: README.md at commit e0a1b26

Tools

0
SourceWeft has not started this package in its sandbox yet, so its tools are not known.

Version history

1
  1. v0.3.1LatestSep 16, 2026

Score

50% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 days25 pts

    Keep the endpoint answering the MCP handshake; SourceWeft probes it every 30 days.

  • Auth declared10 pts

    Declare the credential headers in the registry entry, or keep the endpoint answering the probe so its auth can be classified.

  • Tools listed15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts