Both tools have input schemas with proper JSON Schema structure and type definitions. However, descriptions are in Chinese and lack clarity on when/why to use each tool. Parameter descriptions exist but are minimal. No output schema is documented. Tool names lack clear verb-noun structure ('convert_pdfs' vs 'convert_pdf' is confusing, both do conversion, differ only in scope). Critical issues: no error handling guidance, no pagination support despite batch operations, no security considerations for file operations, and the format_output() function is broken (references undefined 'json' and 'r' variables). The server exposes HTTP file server URLs and concurrency settings as configuration rather than securing them.

• Rename tools to clarify scope: 'convert_pdf_directory' (batch) and 'convert_pdf_file' (single). This eliminates ambiguity and helps LLMs select the right tool.
• Add English descriptions (50-200 chars each) explaining: what the tool does, when to use it (e.g., 'Use this to batch-convert all PDFs in a folder'), and what it returns (e.g., 'Returns a dict mapping PDF paths to lists of image URLs').
• Document output schema explicitly: 'Returns dict[str, list[str]] where keys are PDF paths (relative to PDF_ROOT) and values are lists of image file paths or HTTP URLs (if return_pic_url=true)'.
• Add parameter constraints: 'pdfs_dir must be a valid directory path relative to PDF_ROOT. If invalid, returns error: "Directory not found. Available directories: [list]"'.
• Move HTTP_FILE_SERVER_URL and credentials to environment variables (e.g., MCP_HTTP_FILE_SERVER_URL, MCP_HTTP_FILE_SERVER_TOKEN). Never expose them in config files or logs.

Inferred effective spec: <=2025-11-25.

• Relies on Logging (deprecated) - log to stderr or use OpenTelemetry