Loading ToolBench
A MCP server for searching and downloading academic papers from multiple sources.
Server has 29 tools with consistent naming (verb_noun pattern: search_*, download_*, read_*) and descriptions present for all. However, descriptions are often generic or placeholder-like (e.g., PubMed/Google Scholar tools state 'does NOT support' operations). Parameter schemas are visible and typed, but descriptions lack actionable detail on constraints, formats, and error recovery. Output schemas are not documented. Error handling is minimal, most tools return generic error strings rather than recovery guidance. Three tools (download_pubmed, read_pubmed_paper, download_google_scholar, read_google_scholar_paper) are stubs that explicitly do nothing, wasting tool slots. Sci-Hub integration raises legal/ethical concerns not addressed in descriptions.
Download PDF from arXiv (always free and available).
Download PDF from bioRxiv.
Download PDF from CrossRef.
Google Scholar does NOT support direct PDF downloads.
Download PDF from IACR ePrint Archive.
Download PDF from medRxiv.
PubMed does NOT support direct PDF downloads.
Five tools are non-functional stubs (download_pubmed, read_pubmed_paper, download_google_scholar, read_google_scholar_paper, and implicitly others). Descriptions state 'does NOT support' but tools remain registered, wasting tool slots and confusing LLMs about what is actually available.
Parameter descriptions lack actionable constraints. 'query' params have no guidance on format, length, or special characters. 'save_path' defaults to null with note 'default: ~/paper_downloads' but no validation rules or error messages if path is invalid or unwritable.
No output schemas documented. Tools return 'List[Dict]' or 'str' with no field definitions. LLMs cannot plan downstream calls or extract required IDs (e.g., paper_id from search results) without guessing structure.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 54 | <=2025-11-25 | v2 |
Download PDF from RePEc.
Download PDF from Sci-Hub using DOI.
Download PDF from Semantic Scholar.
Download and extract full text from arXiv paper as Markdown.
Download and extract full text from bioRxiv paper as Markdown.
Download and extract full text from CrossRef paper as Markdown.
Google Scholar does NOT support direct paper reading.
Download and extract full text from IACR paper as Markdown.
Download and extract full text from medRxiv paper as Markdown.
PubMed does NOT support direct paper reading.
Download and extract full text from RePEc paper as Markdown.
Download and extract full text from Sci-Hub paper as Markdown.
Download and extract full text from Semantic Scholar paper as Markdown.
Search preprints on arXiv - major open-access preprint server.
Search biology preprints on bioRxiv.
Search CrossRef database for academic papers.
Search Google Scholar for academic papers.
Search IACR ePrint Archive for cryptography papers.
Search medical preprints on medRxiv.
Search biomedical literature on PubMed (NCBI database).
Search RePEc/IDEAS for economics papers.
Search Semantic Scholar for academic papers.
Error handling is generic. Tools catch exceptions and return 'Error downloading: {str(e)}' or 'Error reading paper: {str(e)}'. No recovery guidance, no categorization (retryable vs fatal), no actionable next steps for LLMs.
Sci-Hub integration (download_scihub, read_scihub_paper) lacks legal/ethical disclaimers. Sci-Hub operates in legal gray zones in many jurisdictions. Descriptions do not warn users or agents about copyright/legal risks.