omni file converter helps convert your PDF,Officle files output to images, documents, pptx and plaintext markdown, etc.
Single tool with critical definitional gaps. Tool name 'pdf_to_image' follows basic verb-noun convention but lacks clarity on behavior. Description is adequate (96 chars, within 10-1024 range) but parameter schema is malformed and incomplete. The function signature shows parameters (file_path, output_format, output_file_name) but the actual convert_pdf_to_images() implementation has a signature mismatch: it declares (input_path, output_format, output_folder, output_file_name), output_folder is NOT exposed as a parameter in the tool definition, creating a critical implementation gap. Parameters lack type clarity in the schema representation. Output schema is documented in docstring as JSON array of dicts with 'output_path' field, but this is not formalized in a structured schema. No error handling guidance for LLM, exceptions return raw error strings without actionable recovery steps.
Convert Your PDF Files to Various Images Formats such as JPG,PNG,JPEG, etc, with High Resolutions.
Parameter-implementation mismatch: tool exposes (file_path, output_format, output_file_name) but convert_pdf_to_images() requires an additional output_folder parameter that is not configurable. The tool will fail at runtime because output_folder is hardcoded nowhere in the tool entry point.
Input schema is partially defined in docstring but not formalized as JSON Schema in the tool registration. The Pydantic decorator uses type hints (str, Optional[str]) which FastMCP should extract, but the actual schema visibility in the source is implicit. Schema lacks enum constraint for output_format despite stating 'Supported values in upper case, include JPG, PNG, JPEG, etc.', LLM can hallucinate invalid formats like 'BMP' or 'PDF'.
Error handling returns raw exception strings ('Error communicating with Bing Image Search API: ...', 'Unexpected error: ...') instead of actionable guidance. LLM receives no hint on whether to retry, check file path, or escalate. Additionally, the Bing Image Search API reference in the error handler is copy-pasted from unrelated code and is misleading.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 39 | <=2025-11-25 | v2 |
| 2026-03-09 | F | 30 | - | v1 |
Output schema is documented only in docstring as informal JSON example. LLM cannot parse this for downstream chaining. No formal structured output schema definition.
Parameter output_file_name description is unclear: 'if not specified, use the same name as input file path file_name.pdf will output file_name_1.jpg, file_name_2.jpg, etc.' This is confusing, does it use the basename of input, or a custom name provided? The logic in convert_pdf_to_images() shows it uses output_file_name if provided, else file_name_clean, but the behavior is not clearly stated.
File path handling has no validation. No check for absolute vs relative paths, no guidance on working directory, and no error message if file does not exist (the function silently returns None on exception, which the tool wrapper will return as a string error). Path traversal attacks possible if agent passes '../../etc/passwd' or absolute paths outside intended directories.