Enterprise Gateway has well-structured schemas and descriptions for most tools, but suffers from inconsistent parameter documentation, missing output schema specifications, and composition issues. 6 of 7 tools have complete input schemas with types and descriptions (baseline: 100% of A+ tools). However, output schemas are not documented anywhere in the source, and several tools lack clarity on return structure. Tool naming follows verb_noun convention (anonymize_, deanonymize_, query_, execute_, get_, security_audit_, dsgvo_pii_). Descriptions range 50-180 chars, within the 10-1024 baseline. Key weakness: execute_code and deanonymize_response lack sufficient parameter validation guidance, and no tool documents pagination or result limits despite query_knowledge accepting top_k.

• Document output schema for each tool. Example for query_knowledge: 'Returns {results: [{id, text, score, project_id}], total_count, has_more}'. This enables LLMs to chain tools and extract required fields.
• Add error recovery guidance to execute_code: 'On timeout (>30s), the container is killed. On syntax error, returns {error, line_number, suggestion}. Retry with corrected code or reduce complexity.'
• Clarify deanonymize_response session lifecycle: 'Session valid for 1 hour after anonymize_prompt call. Returns {original_text, unmapped_placeholders}. If session expired, call anonymize_prompt again.'
• Expand get_okf_concept description: 'Retrieves markdown concept document. Returns {id, title, content, tags, last_updated}. Use for detailed concept exploration; use query_knowledge for search across multiple concepts.'
• Add composition guidance: document which tools are typically called in sequence (e.g., anonymize_prompt → LLM call → deanonymize_response) to help LLMs plan multi-step workflows.