This server has critical definition gaps. Both tools lack parameter descriptions entirely, the rubric requires 100% of A+ tool params to have descriptions. The 'completion' tool has a basic schema with types but no descriptions for 'modelId' or 'prompt'. Tool descriptions are present but generic (under 100 chars), failing to explain WHEN to use each tool or what the LLM should expect. No output schemas are documented. No error handling guidance. No pagination, no idempotent hints, no recovery paths. This is a mock/demo server with minimal production readiness.

• Add descriptions to all parameters. For 'modelId', explain: 'The ID of the model to use (e.g. gpt-4, claude-3-opus, llama-3). Call get_models() first to see available options.' For 'prompt', explain: 'The text prompt to send to the model (1-10000 characters).'
• Document output schemas for both tools. For get_models, specify: returns {models: [{id: string, name: string, provider: string}]}. For completion, specify: returns {modelId: string, completion: string}.
• Expand tool descriptions to 100-200 chars. Example for completion: 'Request a text completion from a specified model. Call get_models() first to see available model IDs. Returns the model's response text.'
• Add parameter constraints. For 'modelId', use an enum of valid model IDs or document the lookup dependency. For 'prompt', specify min/max length and character restrictions.
• Add error handling guidance in the tool description: 'Returns 400 if modelId or prompt is missing. Returns 500 if the model service is unavailable. Retry with exponential backoff.'