Loading ToolBench
MCP server for OpenAlex API - enabling literature review and research landscaping
Strong definition quality with comprehensive parameter documentation and clear tool purposes. Most tools have well-structured input schemas with type definitions, descriptions, and reasonable enums. However, output schemas are entirely undocumented (critical gap), error handling guidance is absent, and tool descriptions lack explicit 'when to use' context for multi-tool scenarios. The server handles 21 read-only tools with consistent verb_noun naming (search_*, get_*, find_*). Parameter documentation is above average (most parameters have descriptions), but missing output schema documentation violates pattern:tool and creates ambiguity for agent planning. Baseline: 72% of tools meet definition standards; output schema documentation gap (-15 points) and absent error recovery guidance (-13 points) prevent higher scores.
Check the quality and ranking of a specific journal or conference across multiple ranking systems (UT Dallas 24, FT50, ABS, etc.). Helps determine if a venue is elite-tier or mid-tier.
Find the top researchers (by h-index) working on a specific topic. Returns experts sorted by h-index with their publication and citation metrics.
Get complete details about a specific author by OpenAlex ID, ORCID, or name. Returns full author profile including h-index, i10-index, citation metrics, all affiliations, and research topics.
Get all works published by a specific author. Includes filtering and sorting options to find an author's most cited or recent publications.
Get papers that cite a specific work. Useful for tracking a paper's influence and finding related work that builds upon it.
Get complete details about a specific institution by OpenAlex ID or ROR ID. Returns institution profile including research metrics, geographic location, and associated institutions.
Output schemas completely undocumented. No return type definitions for any of the 21 tools. LLMs cannot predict response structure, forcing runtime discovery and risking malformed downstream calls.
No error handling guidance. Tools lack descriptions of failure modes, recovery strategies, or actionable error messages. Agents cannot self-correct on invalid inputs or API failures.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-21 | C | 62 | <=2025-11-25 | v2 |
| 2026-03-09 | F | 47 | - | v1 |
Get publication and citation statistics for a work, author, institution, or venue. Returns year-by-year and cumulative metrics.
Find works related to a given work based on shared topics, citations, and references. Useful for discovering similar papers in a research area.
Get publication trends over time for a topic, field, or institution. Shows number of publications per year to visualize research growth or decline.
Get complete details about a specific publication venue (journal or conference) by OpenAlex ID or ISSN. Returns venue metrics, host organization, and research scope.
Get COMPLETE details about a specific work by OpenAlex ID or DOI. Unlike search results which are summarized, this returns ALL information including: complete author list (first, middle, and last authors with positions, institutions, ORCID, corresponding author flags), full abstract (reconstructed), all topics, complete bibliographic data, funding/grants, keywords, and reference lists. Use this when you need detailed information about a specific paper, especially for identifying PIs (often last author) or corresponding authors.
Get the list of references (bibliography) cited by a specific work. Useful for understanding the foundational literature a paper builds upon.
List all available institution group presets (e.g., Ivy League, Top US, Harvard/Stanford/MIT). Use these preset names with search_works or other search tools via the institution_group parameter to filter by elite institution groupings.
List all available journal and conference presets with descriptions. Use the preset names returned here with search_in_journal_list to search within curated top-tier venue lists (e.g., UT Dallas 24, FT50, ABS rankings, Nature/Science family).
Search for academic authors by name. Returns author profiles with publication counts, citation metrics (h-index, i10-index), and current institutional affiliations. Useful for finding domain experts or co-authors.
Search for works within specific research topics or domains. Use this to explore literature in a particular field or subfield. Supports venue filtering to restrict results to top journals/conferences.
Search works within a curated preset journal/conference list (e.g., UT Dallas 24, FT50, ABS 4*, Nature/Science). Useful for finding papers in top-tier venues only. Use list_journal_presets to see all available preset lists.
Search for academic institutions by name or country. Returns institution profiles with publication counts, citation metrics, and geographic information.
Search for journals, conferences, or other publication venues by name or ISSN. Returns venue metadata including impact metrics, publication type, and scope.
Search for research topics or domains in OpenAlex. Returns topic metadata with publication and citation counts, helping you explore literature landscapes.
Search scholarly works with advanced filtering. Supports Boolean operators, year ranges, citation thresholds, venue/journal filtering (source_name, source_issn, source_id), and institution filtering (author_institution, institution_group). The most flexible search tool. Use search_in_journal_list for preset journal lists like UTD24 or FT50.
Ambiguous tool selection guidance. 11 tools follow search_* or get_* patterns on the same domain (authors, sources, institutions, topics). Descriptions do not clearly differentiate when to use search_authors vs find_experts, or search_topics vs search_by_topic. LLMs may hallucinate tool names or waste reasoning cycles.
Missing parameter defaults and constraints. Many parameters (sort, per_page, page) lack guidance on defaults or reasonable ranges. Descriptions mention 'default: 10' inline but do not validate against absurd values. LLMs can pass per_page=1000 and exhaust context windows.
Missing mutual exclusivity documentation. search_works accepts both exact_phrase and search_field but notes state 'Cannot be combined with exact_phrase.' No validation or error guidance if both are passed.