MCP server for managing GPU computing instances, SSH keys, API keys, and billing information on NebulaBlock platform
NebulaBlock exposes 17 tools with basic naming and descriptions, but lacks critical depth. Tool names follow verb_noun convention (get_, list_, create_, delete_) which is good. However, descriptions are inconsistent: some tools have multi-line docstrings (e.g., delete_gpu_instance, create_gpu_instance) while others have single-line summaries. Parameter descriptions exist but are minimal (e.g., 'Instance ID' for id param). No output schemas are documented, LLMs cannot predict response structure. Error handling is absent: raw HTTP exceptions will propagate. No parameter validation, enums, or constraints visible. Pagination parameters (limit, offset) appear in 4 tools but lack min/max bounds. The server accepts credentials via environment variables (good), but no permission gates or audit logging visible.
Create GPU Instances. Create an instance with the specified custom configuration and features provided in the request body.
Creates an SSH key for use in your instances.
Delete GPU Instances. Permanently delete an instance by specifying the instance ID in the path to delete the selected instance.
Deletes a specified SSH key by including the ID of the SSH key in the endpoint path.
Get available computing products
Retrieve the user's transaction history. Retrieve your credit payment history for different products.
No output schemas documented. LLMs cannot predict response structure, forcing them to guess what fields are available for downstream tool calls or data extraction.
Pagination parameters (limit, offset) lack min/max bounds. No guidance on default values or maximum safe limits. LLMs may pass absurd values (limit=999999) causing timeouts or API errors.
No error handling or recovery guidance. Raw HTTP exceptions (raise_for_status) will propagate as stack traces. LLMs receive no actionable error messages or retry hints.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 54 | 2026-07-28+ | v2 |
Get user's current credit balance
Get detailed information about a specific GPU instance
Get user's GPU instances with optional pagination
List API keys
List Available OS Images. Return a list of all available operating system images, including details about each image's version and driver, if applicable.
List deleted GPU instances
List SSH keys
List user's invoices with optional pagination
Reboot GPU Instances. Initiate a reboot of an instance. Provide the instance ID in the path to reboot the specified instance.
Start GPU Instances. Initiate the startup of an instance. Provide the instance ID in the path to start the specified instance.
Stop GPU Instances. Shut down an instance. Provide the instance ID in the path to initiate the shutdown process for that instance.
Destructive operations (delete_gpu_instance, delete_ssh_key) lack confirmation or dry-run support. Agents can permanently delete resources without safeguards.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint) visible. Clients cannot distinguish safe reads from destructive writes without parsing descriptions.