Promptql MCP Server
Natural language queries for Hasura PromptQL data agents via MCP.
Data last scanned today · Reviewed today
Overview
This server acts as a bridge, allowing AI assistants to interface directly with Hasura's PromptQL data agents. By utilizing natural language processing, users can extract insights, perform data analysis, and visualize enterprise data through formatted table artifacts directly within their AI environment. It supports multiple authentication configurations to accommodate both public and private Hasura DDN deployments. Users can authenticate using either standard auth tokens or specific headers required for internal private endpoints, providing flexibility for varied infrastructure setups.
Our verdict
With a security score of 29/100, this project carries a high risk level. The score reflects that the server is community-maintained rather than vendor-official and has not undergone an editorial security review. Furthermore, the repository shows limited activity, with the last commit occurring over 200 days ago. It is recommended only for experimental or non-production environments where security auditing and maintenance updates can be managed internally.
- Supports both public and private Hasura DDN authentication modes
- Provides native support for rendered table artifacts
- Features zero direct dependencies to reduce supply-chain surface area
- Licensed under the permissive MIT license
- High risk score of 29/100 due to community-maintained status
- Last code commit occurred more than 200 days ago
- Lacks external editorial security review of read-only mode
- Explicitly marked as unstable and limited to PromptQL v1
Tools
| Tool | Description | Risk |
|---|---|---|
| start_thread | Start a new conversation thread with an initial message and wait for completion (returns thread_id, interaction_id, and response) | medium |
| start_thread_without_polling | Start a new conversation thread without waiting for completion (returns thread_id and interaction_id immediately) | medium |
| continue_thread | Continue an existing thread with a new message (maintains conversation context) | medium |
| get_thread_status | Check the status of a thread (processing/complete) using GET /threads/v2/{thread_id} | low |
| cancel_thread | Cancel the processing of the latest interaction in a thread | medium |
| setup_config | Configure PromptQL API key, playground URL, DDN Auth Token, and authentication mode (public/private) | medium |
| check_config | Verify the current configuration status including authentication mode | low |
Compatibility
| Client | Local | Docker | Remote | Read-only |
|---|---|---|---|---|
| Claude Desktop | ||||
| Cursor | ||||
| VS Code | ||||
| Windsurf | ||||
| ChatGPT |
Frequently asked questions
›What are the authentication differences between public and private DDN deployments?
Public mode uses the Auth-Token header for connectivity, while private mode uses the x-hasura-ddn-token header for increased security with internal endpoints.
›What are the minimum system requirements for the PromptQL MCP server?
The server requires Python 3.10 or higher, along with a Hasura PromptQL project that includes an API key, playground URL, and DDN Auth Token.
›Can I configure my credentials using environment variables instead of the setup command?
Yes, you can export PROMPTQL_API_KEY, PROMPTQL_PLAYGROUND_URL, PROMPTQL_AUTH_TOKEN, and PROMPTQL_AUTH_MODE as environment variables.
›Is this version of the server compatible with PromptQL v2?
No, this version of the server is currently marked as unstable and supports only PromptQL v1, with an alpha release for v2 expected soon.
›How do I ensure the Claude Desktop configuration uses my virtual environment?
In your Claude Desktop config file, specify the full path to the Python executable located within your virtual environment's bin folder.
Alternatives
Badge
Maintain this server? Add the live badge to your README.