Single tool with minimal viable definition. Tool is explicitly registered via @mcp.tool decorator in main.py with a docstring, but the schema is entirely empty (no input parameters), description is extremely brief (9 words), and there is no output schema documentation. The tool does one thing (generate random number) which is good for composition, but lacks the rigor expected of production tools. No parameter descriptions, no input constraints, no output structure documentation, no error handling guidance.

• Rename tool from 'random_number' to 'generate_random_number' to follow verb_noun convention and match the action verb baseline (90% of A+ tools start with action verbs like get, create, search, update).
• Expand description to 50 - 150 characters explaining WHAT (generates random int), WHEN to use (when needing non-deterministic testing or randomization), and WHAT it returns (integer 1 - 100). Target: 'Generate a random integer between 1 and 100 inclusive. Useful for sampling, randomization, or testing variability. Returns a single int.'
• Document the output schema explicitly. Add a return type annotation and docstring: '@mcp.tool returns int: A random integer in the range [1, 100].'
• Add input parameter if range should be configurable (e.g. min, max parameters with defaults). If truly parameter-less, state '(Takes no parameters)' in the description to prevent LLM confusion about whether args are expected.
• Add error handling: wrap random.randint() in a try/except, return structured error with recovery hint if generation fails (unlikely but good practice).