Loading ToolBench
MCP server for searching and downloading academic papers from multiple research sources
Research Master has 18 well-named tools with consistent verb_noun patterns (search_papers, get_paper, download_paper, etc.). All tools have descriptions (avg ~80 chars, within baseline 34-392 range). Input schemas are present and properly typed with JSON Schema. However, output schemas are NOT documented, critical gap for LLM planning. Parameter descriptions are minimal (often 1-2 sentences). Error handling guidance is absent. No tool annotations (readOnlyHint/destructiveHint). Several tools accept free-form strings where enums would be better (e.g., 'source' param, 'format' param in export_papers). Missing dependency hints and recovery guidance.
Get author profile information and publication history.
Get metadata for multiple papers in a single request.
Get citation graph data for a paper.
Deduplicate papers from search results across multiple sources.
Download a paper PDF to your local filesystem. Source is auto-detected from paper ID format.
Export papers to various formats (BibTeX, CSV, JSON, RIS).
Get papers that cite a specific paper. Prefers Semantic Scholar for best results.
Output schemas not documented. LLMs cannot plan downstream tool calls or extract required fields (e.g., paper_id for chaining). Critical for multi-step research workflows.
'source' parameter accepts free-form strings (e.g., 'arxiv', 'semantic', 'pubmed') without enum constraint. LLMs will hallucinate invalid source names. Should be enum: ['arxiv', 'semantic', 'pubmed', 'biorxiv', ...].
'format' parameter in export_papers lacks enum constraint. Should be enum: ['bibtex', 'csv', 'json', 'ris']. Free-form strings invite invalid values.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 66 | <=2025-11-25 | v2 |
Get detailed metadata for a specific paper. Source is auto-detected from paper ID format.
Get papers referenced by a specific paper. Prefers Semantic Scholar for best results.
Get papers related to a specific paper.
List all available research sources and their capabilities.
Look up a paper by its DOI (Digital Object Identifier).
Extract and return the full text content from a paper PDF. Source is auto-detected from paper ID format. Requires poppler to be installed.
Get paper recommendations based on a given paper or query.
Search for papers by author across research sources
Search for papers across multiple available research sources
Search the web for papers and academic content.
Write paper metadata to a file.
No error handling guidance. Tools lack recovery hints (e.g., 'Paper not found. Try search_papers() with a partial title.'). Agents cannot self-correct on failures.
No tool annotations (readOnlyHint, destructiveHint). download_paper and write_paper are WRITE operations but lack destructiveHint. Agents cannot distinguish safe from risky tools.