Single tool 'evaluate_draft' has a basic schema but critical gaps in parameter descriptions and output documentation. The 'criteria' parameter is typed as 'object' with no description of its structure, required fields, or constraints. Parameter descriptions are missing entirely (text and criteria lack descriptions in the schema). Output schema is undocumented, LLMs cannot predict the response structure. Tool description is minimal (11 words, 60 chars) and lacks WHEN/WHY guidance. No error handling guidance. Naming is acceptable (verb_noun form) but the tool lacks the rigor expected for production use.

• Add detailed description to 'criteria' parameter: 'Dictionary mapping criterion names to evaluation specs. Each spec must contain: weight (float, default 1.0), min_words (int, minimum word count), must_contain (list of strings, required keywords). Example: {"depth": {"weight": 1.5, "min_words": 15}, "terminology": {"weight": 1.0, "must_contain": ["latency", "throughput"]}}.'
• Add description to 'text' parameter: 'The text draft to evaluate (string, 1-10000 characters). Can be any natural language content: essay, email, code comment, etc.'
• Document the output schema in the tool description or as a separate schema field: 'Returns {overall_score (0-100), status (PASS|NEEDS_REFINEMENT), criterion_scores (dict of criterion→score), critique_points (list of actionable feedback strings), word_count (int)}.'
• Expand tool description to 100-150 characters: 'Evaluate text drafts against multi-criteria rubrics. Scores each criterion by word count and required keywords, returns weighted overall score and targeted critique points. Use to validate content quality before publishing.'
• Add error handling: validate criteria structure and return clear errors like 'Invalid criteria: missing required field "weight" in criterion "depth"' instead of silent failures.