Loading ToolBench
MCP server exposing pdf-inspector tools to coding agents
16 tools with basic schemas and descriptions, but significant gaps in parameter documentation and output schema clarity. Most tools accept a single 'path' parameter with minimal constraint documentation. Descriptions are present but generic (avg ~60 chars, well below the 194-char baseline). No parameter descriptions for complex inputs like 'regions' array. Output schemas are not documented, LLMs cannot predict return structure. Error handling is absent from descriptions. Tools like compare_line_items have 6 parameters but lack per-parameter descriptions and validation guidance. No tool annotations (readOnlyHint, etc.). Composition is reasonable (single responsibility per tool), but parameter naming could be more explicit (e.g., 'regions' lacks format specification).
Analyze the layout of a PDF document
Classify multiple PDF documents in batch
Classify a local document without invoking a conversion parser
Classify a PDF document
Compare a return line against a source document line
Return the generic document contract for a given format
Convert a document to Markdown using the generic document route
Output schemas not documented. LLMs cannot predict return structure (fields, types, nesting). Tools like classify_pdf, pdf_to_markdown, and analyze_layout return unspecified JSON, agents must guess what fields exist.
Complex parameters lack descriptions. 'regions' in extract_text_regions and extract_table_regions is documented as 'Regions to extract from, specified as (page, rects) pairs' but lacks format specification, type details, and examples. LLMs cannot construct valid region arrays.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 47 | <=2025-11-25 | v2 |
Extract table data from specified regions of a PDF
Extract text from specified regions of a PDF
Identify tax form type in a PDF document
List available tax review demo packages
Parse Internal Revenue Code sections from a document
Convert a PDF document to Markdown
Render a review memo for a tax package
Review a tax package and return analysis
Split an SEC filing document into sections
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). All tools are READ_ONLY but this is not declared in the tool definition, agents cannot infer safety properties from the schema alone.
Descriptions are generic and brief (avg ~60 chars vs 194-char baseline). Most lack WHEN to use, WHAT it returns, and dependencies. E.g., 'Classify a PDF document' does not explain what classification means, what categories are returned, or when to call it vs identify_tax_form.
No error handling guidance in descriptions. Tools do not document what happens on invalid paths, unsupported file formats, or parsing failures. Agents have no recovery path.