Loading ToolBench

Reading the current corpus…

amira-mcp-server MCP server · F · ToolBench

amira-mcp-server

github.com/AM-Digital-Research-Environment/amira-mcp-server

Read-only access to AMIRA, the research-data platform of the Africa Multiple Cluster of Excellence at the University of Bayreuth (Omeka S).

2026-07-28+STDIO26 tools

41/100

Definition Quality38

Protocol Readiness50

SupportabilityNot measured-

Spec posture

Inferred effective spec 2026-07-28+·current baseline 2026-07-28

Static source inference · medium confidence · evidence: stateless requests, Apps extension

Current-spec patterns detected

Summary

AMIRA exposes 26 read-only tools with minimal descriptions and no visible input/output schemas in the provided source. Tool names follow verb_noun convention (get_*, search_*, list_*), which is good. However, descriptions are generic (e.g., 'Search research items in AMIRA'), parameter schemas are not visible in the code excerpts, and no error handling guidance is documented. The server is STDIO-only, which caps protocol readiness. Most tools appear to be inferred from scripts/weigh.mjs rather than explicitly registered with full schemas in src/tools/register.ts, triggering the inference penalty (cap at 50 per tool). No tool annotations (readOnlyHint, destructiveHint, idempotentHint) are evident despite all tools being read-only.

Tools (26)

fetchread onlysource verified35/100

OpenAI-compatible fetch tool for ChatGPT (HTTP transport only)

find_relatedread onlysource verified42/100

Find entities related to a subject, place, person or project

get_collection_overviewread only40/100

Get collection-level statistics and metadata about AMIRA

get_institutionread onlysource verified40/100

Get detailed information about a specific institution

get_personread onlysource verified40/100

Get detailed information about a specific person

get_podcastread onlysource verified40/100

Get detailed information about a specific podcast

get_projectread onlysource verified40/100

Get detailed information about a specific project

get_publicationread only

Issues (5)

CRITICALtool-description, constrained-input ↗

No input/output schemas visible in source code. All 26 tools lack documented parameter types, constraints, and return structures. Cannot verify LLM can infer valid inputs or parse results.

get_collection_overviewsearch_research_itemssearch_personssearch_projectssearch_publicationssearch_podcastssearch_videossearch_placessearch_subjectssearch_languagesget_research_itemget_personget_projectget_research_sectionget_institutionget_publicationget_podcastget_videolist_yearslist_research_sectionslist_institutionslist_categorieslist_publication_facetsfind_relatedsearchfetch

HIGHtool-description ↗

Descriptions are generic and under 50 characters for most tools (e.g., 'Search research items in AMIRA'). Lack context on WHEN to use each tool, dependencies, or expected result structure. LLMs cannot distinguish between similar search_* tools.

Recommendations

Add explicit JSON Schema definitions for all 26 tools in src/tools/register.ts. Include input parameter types (string, number, enum), constraints (min/max, pattern), and descriptions. Document output schema with field types and descriptions.
Expand tool descriptions to 100 - 200 characters. For each search_* tool, explain: what entity type it searches, what fields are indexed, when to use it vs. similar tools, and what the response structure contains. Example: 'Search the AMIRA research items database by title, keywords, or metadata. Returns up to 50 items with ID, title, date, and related project/person IDs. Use get_research_item to fetch full details.'
Add readOnlyHint: true to all 26 tool definitions to signal to clients that results are cacheable and calls are safe to retry without side effects.
Document parameter constraints in descriptions. For search tools, specify: max query length, supported filter fields, result limit, and pagination (offset/limit or cursor). Example: 'query (string, 1 - 500 chars): Search term. limit (integer, 1 - 100, default 20): Max results. offset (integer, default 0): Pagination offset.'
Add error handling guidance to tool descriptions. Example: 'Returns 404 if the research item ID does not exist. Returns 400 if the query exceeds 500 characters. Returns 503 if the AMIRA service is temporarily unavailable, retry after 30 seconds.'

Spec posture evidence

Inferred effective spec: 2026-07-28+.

Current-spec patterns present: stateless request handling

Score history

Overall score trend

First recorded score · v2 rubric

41/100

Scored	Grade	Overall	Spec posture	Rubric
2026-09-23	F	41	2026-07-28+	v2