tigris

The Tigris MCP Server exposes Tigris bucket and object management capabilities through the MCP specification, enabling AI agents to interact with your Tigris storage as a context provider. Once running, agents can issue prompts like listing buckets, creating or deleting buckets, listing objects, uploading or modifying objects, creating folders, generating shareable links, and more, all mediated by the MCP server. This makes it easy to integrate Tigris storage into AI editor workflows or agent-driven data pipelines. The server supports common authentication patterns via AWS-style credentials, and can be used with either NPX (local Node.js runtime) or Docker, with Docker recommended for sandboxing and isolation. To connect, provide your AWS credentials and endpoint, and point your MCP client at the running tigris-mcp-server instance. The server will handle S3-compatible requests against Tigris, exposing bucket and object operations to your agents.

Prerequisites:

• Docker installed or Node.js (for NPX usage)
• Access to a Tigris account with an S3-compatible endpoint and credentials (AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY)

Installation options:

Option 1: NPX (local Node.js runtime)

1. Ensure Node.js and npm are installed.
2. Run the MCP server directly with NPX:

npx -y @tigrisdata/tigris-mcp-server run

Option 2: Docker (recommended for sandboxing)

1. Ensure Docker Engine is installed.
2. Run the MCP server in a container using the provided image:

docker run -e AWS_ACCESS_KEY_ID -e AWS_SECRET_ACCESS_KEY -e AWS_ENDPOINT_URL_S3=https://fly.storage.tigris.dev -i --rm quay.io/tigrisdata/tigris-mcp-server:latest

Option 3: VS Code / VS Code Insiders one-click install (as described in the README)

• Use the provided install configuration snippets (they set command to npx and include AWS credentials and endpoint) and install via the VS Code MCP server integration.

Notes:

• Replace placeholder credentials with your actual AWS keys or use AWS profiles if you prefer.
• When using Docker, the config often binds a sandboxed path per user for security.

Tips:

• If you already use AWS CLI, you can leverage AWS_PROFILE by setting USE_AWS_PROFILES=true and AWS_PROFILE=default along with AWS_ENDPOINT_URL_S3.
• The AWS_ENDPOINT_URL_S3 value should point to the Tigris S3-compatible endpoint (e.g., https://fly.storage.tigris.dev).
• For client configurations outside of NPX, you can adapt to Docker-based execution by following the alternate JSON blocks shown in the README (with docker command and environment mappings).
• Ensure your credentials are kept secure and not committed to any public repository. Use environment variables or secret managers where possible.
• If you encounter sandboxing issues in NPX mode, prefer Docker which provides better isolation.