Six tools with minimal descriptions (10-40 chars, well below 50-200 char baseline for LLM optimization). All tools have input schemas with typed parameters and required fields, but descriptions are cryptic abbreviations ('discover endpoints', 'full param + response spec') that do not explain WHEN to use each tool, WHAT it returns, or dependencies. No output schemas documented. No error handling guidance. Tool naming is verb-noun compliant (list_, describe_, call_) but descriptions fail to disambiguate the three tools per surface or explain the write-gate mechanism. Parameters lack descriptions for 'reads_only', 'limit', 'path_params', 'params', LLMs cannot infer what these control. No mention of pagination, result limits, or response structure. Security: call_endpoint tools are WRITE-gated but no description explains the gate logic or what 'validate' entails.

• Expand tool descriptions to 50-200 chars. Example: 'pve_list_endpoints: Discover available Proxmox VE API endpoints. Returns operation_id, method, path, and description for each. Use this first to find the endpoint you need, then call pve_describe_endpoint to see full parameter specs.'
• Add descriptions to all parameters. Example: 'reads_only (boolean): Filter to read-only endpoints (GET/HEAD only). Omit or false to include write endpoints (POST/PUT/DELETE).' 'limit (integer): Maximum endpoints to return (1-1000, default 50). Use pagination to fetch large catalogs.'
• Document output schemas for each tool. Example: 'pve_list_endpoints returns {endpoints: [{operation_id, method, path, description, destructive}], total, next_cursor}'. This enables LLMs to chain calls (list → describe → call).
• Add error handling guidance to call_endpoint. Example: 'Validates operation_id, path_params, and params against the endpoint spec. Returns 400 with constraint violations (e.g., "Invalid param: status must be one of: running, stopped"). Returns 403 if the operation requires write permission and credentials lack it.'
• Clarify the three-tool workflow in descriptions. Example: '(1) pve_list_endpoints to discover endpoints, (2) pve_describe_endpoint to see full parameter specs and response schema, (3) pve_call_endpoint to execute. All three are required for safe, informed API calls.'