Two well-defined tools with complete schemas, clear descriptions, and proper type annotations. Both tools are read-only, idempotent, and correctly annotated. Naming follows verb_noun convention (check_statements, describe_schema). Descriptions are substantive (150 - 200 chars) and explain WHAT, WHEN, and WHY. Input schemas are fully typed with descriptions for all parameters. Error handling in check_statements returns actionable hints. Main gaps: no output schema documentation in docstrings, no pagination (not applicable here), and parameter descriptions could be more prescriptive about format/constraints.

• Add output schema documentation to both tool descriptions. Example: 'Returns: {verdict: bool, failing_checks: [{name: str, expected: float, actual: float}], counts: {passed: int, failed: int}, summary: str}'.
• Specify numeric parameter ranges in descriptions: 'abs_tol (float, default 0.5): Absolute tolerance for comparisons; must be >= 0, typical range 0.0001 - 1.0.'
• Explicitly state the discovery pattern: 'Call describe_schema() first to learn the canonical field names and sign convention, then structure your input accordingly.'
• Enhance error responses with error classification: return {error: str, category: 'user_input_error'|'missing_data'|'internal_error', hint: str} to guide LLM recovery.
• Document the expected structure of nested dicts (periods → field names → values) with a brief example in the description or a link to the schema tool.