Loading ToolBench
MCP server for translating PDF and Word documents with support for mathematical content. Provides PDF/DOCX translation capabilities using various translation services.
Single tool server with moderate definition quality. Tool has a schema with typed parameters and descriptions, but lacks critical patterns for production use. The description is adequate (145 chars) but generic. Parameters lack enums, validation ranges, and error handling guidance. No idempotency, confirmation, or dry-run patterns. The tool performs file I/O with WRITE risk but no output schema is documented. Error messages are unstructured strings rather than actionable recovery guidance. Definition falls short of A+ baseline (which requires 50-200 char descriptions, full parameter validation, documented output schemas, and error guidance).
translate given pdf or word document. Argument `file` is absolute path of input pdf/doc/docx, `lang_in` and `lang_out` is translate from and to language, and should be like google translate lang_code. `lang_in` can be `auto` if you can't determine input language.
No output schema documented. Tool returns a free-text string summary rather than a structured object with documented fields. LLMs cannot reliably extract the output file paths or status.
Missing enum constraints on language parameters. 'lang_in' and 'lang_out' are free-form strings with only a note that they should be 'like google translate lang_code'. No enum list provided, inviting hallucinated language codes (e.g., 'en_US' vs 'en', 'zh_CN' vs 'zh').
No error handling or recovery guidance. The tool can fail (file not found, invalid language, translation service unavailable, out of disk space) but error responses are raw strings with no guidance on what the LLM should do next.
Destructive operation (file I/O: creates two output PDFs) with no confirmation or dry-run pattern. If the agent is testing or iterating, this repeatedly overwrites output files without user awareness.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | C | 60 | <=2025-11-25 | v2 |
| 2026-03-09 | F | 31 | - | v1 |
Parameter 'file' requires absolute path but does not validate path traversal or access control. Agents could be tricked into reading/writing sensitive files outside intended directories.
Description mentions 'lang_in' can be 'auto' but does not explain what 'auto' triggers (automatic language detection) or its success rate. LLMs may over-rely on it.
No documented handling of supported file formats. Description says 'pdf or word document' but actual support matrix (docx, doc, others) is unclear from the description alone.