Loading ToolBench
MCP server for parsing interleaved multimodal documents using vision language models, extracting LaTeX equations, charts, and generating hierarchical markdown from document images
Critical failures across all dimensions. The single tool 'execute_skill_action' has a reasonable name and basic schema, but the implementation is fundamentally broken: mcp_server.py contains only a stub that prints JSON and exits, it does not implement the MCP protocol at all. No actual tool invocation, no request/response handling, no error handling. The client.py file is a mock that returns hardcoded responses, not a real MCP server. Tool description is present but generic (89 chars). Input schema exists with two parameters (page_image_url, extract_latex_equations) but lacks output schema documentation. No error handling, no validation, no security considerations. This is a non-functional skeleton, not a production MCP server.
Execute the interleaved multimodal document vision language parser skill to parse document pages and extract structured content
MCP server implementation is a non-functional stub. mcp_server.py prints a JSON status message and exits; it does not implement the MCP protocol, handle tool calls, or process requests. The server cannot receive or respond to tool invocations.
No output schema documented. The tool description does not specify what fields the response contains. Clients and LLMs cannot plan downstream operations or extract required data (e.g., parser_page_id, structured_document_url visible in mock but not declared).
Tool description is generic and lacks actionable context. '89 chars: 'Execute the interleaved multimodal document vision language parser skill to parse document pages and extract structured content', does not explain WHEN to use it, what prerequisites exist, or what the response structure is.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 34 | 2026-07-28+ | v2 |
No error handling or recovery guidance. If the image URL is invalid, the service is unavailable, or LaTeX extraction fails, the tool provides no actionable error message or retry strategy.
Input parameters lack validation constraints. 'page_image_url' accepts any string; no format validation (URL pattern), no timeout, no size limits. 'extract_latex_equations' is boolean but no description of side effects or performance impact.