This server has severe definition quality issues. Tool names do not follow verb_noun conventions (recommend_tool is acceptable but dynamic_pattern_tools is vague). Descriptions are present but lack critical detail about when/why to use each tool and what they return. Input schemas are minimal, only a single 'input' string parameter with no validation, constraints, or format guidance. Output schemas are completely undocumented. The dynamic_pattern_tools description is generic and does not explain the pattern selection logic, constraints, or expected output structure. Error handling is minimal, no recovery guidance, no error classification, no actionable error messages. The recommend_tool hardcodes 'summarize' as a fallback and has commented-out Perplexity integration, indicating incomplete implementation. No per-tool security, composition, or chaining guidance. Average baseline for production tools: descriptions 194 chars (this server: recommend_tool ~75 chars, dynamic_pattern_tools ~150 chars, adequate length but missing actionable detail). Parameters lack enums, format constraints, or range validation. No documented output schemas. This is a D-/F+ server.

• Rename 'dynamic_pattern_tools' to action-verb form. Each pattern should be exposed with its own tool name (e.g., 'extract_text', 'summarize_document', 'generate_questions'). Dynamically register each as a separate tool with a verb_noun name and domain-specific description.
• Document output schema for each tool. Example: 'recommend_tool returns {"recommended_pattern": string, "confidence": number (0-1), "reason": string, "available_patterns": string[]}'. Example: 'Each pattern tool returns {"pattern": string, "result": string, "metadata": {"execution_time_ms": number}}'.
• Add parameter validation and constraints. Replace bare 'input' with descriptive guidance: 'Input text for the pattern (1-10000 chars, plaintext or markdown).' Add enums if pattern-specific inputs are expected.
• Replace hardcoded 'summarize' fallback in recommend_tool with a real error recovery: 'Unable to recommend a pattern. Available patterns: [list]. Try again with a more specific task description.' Log the original error for debugging.
• Expand tool descriptions to include domain context and selection criteria. Example for a pattern tool: 'Extract key entities (people, places, organizations) from unstructured text using NLP-based pattern matching. Use when you need to identify named entities in user input or documents.' (vs. vague 'Execute the extract Fabric pattern').
• Add per-parameter descriptions that guide LLM input. Example: 'input: The text to process (required, plaintext or markdown, 1-10000 characters). Include the full context you want the pattern to analyze.'