Both tools have minimal descriptions (under 100 chars) and lack parameter descriptions entirely. Input schemas are present but sparse: 'diff' and 'path' are typed as strings with brief descriptions, but lack constraints (format, length, examples of valid input). No output schema is documented. Tools return JSON strings rather than structured objects. Error handling is present (billing checks) but responses are JSON-encoded strings, not proper error objects. No parameter validation guidance. Naming is reasonable (review_code, check_quality) but descriptions don't explain WHEN to use each tool or what distinguishes them.

• Expand tool descriptions to 100-200 chars. Explain WHAT each tool does, WHEN to use it (e.g., 'Use review_code for diffs; use check_quality for file paths'), and what it returns.
• Add parameter descriptions with format and constraint guidance: 'diff (str): Unified diff format or file content (max 100KB). Include file paths and line numbers for context.' and 'path (str): Absolute or relative file path (max 500 chars). Must be readable by the server.'
• Document output schema in docstrings. E.g., 'Returns: {"issues": [{"type": "bug|security|style", "line": int, "message": str}], "credits_remaining": int}'.
• Return structured error objects instead of JSON strings: raise ValueError('Out of credits...') and let FastMCP serialize it, or return {"error": {...}, "type": "billing_error"} with a clear error type field.
• Add input validation: check diff length, validate path is not absolute/traversal, return actionable errors like 'diff exceeds 100KB limit' instead of silent failures.