Server has 9 tools with basic schemas and descriptions in Chinese. All tools have input schemas with type definitions and parameter descriptions, meeting minimum requirements. However, descriptions are terse (avg ~60 chars, well below the 194-char baseline for A+ tools), lack WHEN/WHY context, and omit error recovery guidance. No output schemas documented. Tools lack error handling patterns, confirmation steps for destructive ops (qz_auth_login, qz_set_cookie), and actionable error messages. Naming is verb-noun compliant but generic (list_*, get_*). No pagination guidance despite list tools. Security concern: credentials passed as parameters (username/password in qz_auth_login, cookie in qz_set_cookie) violates secret-injection pattern.

• Move credential handling to server-side secret injection: read cookie from environment variable or vault, not from tool parameters. Implement qz_auth_login as a server initialization step, not a tool.
• Expand tool descriptions to 100-200 chars with WHEN/WHY context. E.g., 'qz_list_jobs: List all jobs in a workspace. Use this to discover available jobs before calling qz_get_job for details. Returns job_id, name, status, and compute resources.'
• Document output schemas in tool descriptions or as a separate schema field. E.g., 'Returns: {job_id, name, workspace_id, status_raw, status_normalized, status_family, is_terminal, is_active, created_at, finished_at, url}'.
• Add error recovery guidance to descriptions. E.g., 'If cookie is not set, call qz_auth_login first. If workspace not found, call qz_list_workspaces to see available options.'
• Add pagination guidance to list tools: 'Supports limit parameter (default 100). For large result sets, use limit and offset to paginate. Returns total count for planning.'