Loading ToolBench

Reading the current corpus…

ModelContextProtocol MCP server · F · ToolBench

ModelContextProtocol

github.com/miguelxp12/ModelContextProtocol

Implementación básica del Model Context Protocol (MCP) de Anthropic con capacidades de búsqueda y recuperación de documentos

2026-07-28+STREAMABLE_HTTP5 tools

40/100

Definition Quality47

Protocol Readiness35

Supportability33

Spec posture

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

Static source inference · medium confidence · evidence: Streamable HTTP

Current-spec patterns detected

Summary

This MCP server has significant structural and quality issues. Tool definitions exist but lack proper schema documentation and comprehensive descriptions. Five tools are registered (search, retrieve, list, update, ping), but input/output schemas are not formally documented in the source code. The server uses a custom HTTP endpoint (/mcp) without following MCP's current stateless request model or structured tool metadata patterns. Descriptions are present in Spanish but vary widely in quality and detail. No input validation guidance, error handling strategies, or confirmation patterns for destructive operations are evident. The 'ping' tool violates current MCP spec (ping was removed). Tool composition is reasonable (single concerns mostly), but parameter constraints and type safety are weak.

Tools (5)

listread onlysource verified45/100

Listar todos los documentos. Retorna una lista de todos los documentos disponibles con su ID y título.

pingread onlysource verified37/100

Verificar la conexión con el servidor MCP. Retorna un pong con timestamp y versión.

retrieveread onlysource verified53/100

Recuperar documento por ID. Obtiene el contenido completo de un documento específico por su identificador.

searchread onlysource verified53/100

Búsqueda básica en documentos. Busca documentos cuyo título o contenido incluya la consulta proporcionada.

updatewritesource verified47/100

Actualizar un documento (simulado). Modifica el contenido de un documento existente identificado por su ID.

Issues (7)

CRITICALtool-schema ↗

Tool schemas not formally documented in source code. Input/output parameters exist in a basic object form but lack complete JSON Schema definitions with type constraints, required fields, and structured validation metadata.

searchretrievelistupdateping

HIGHspec-compliance ↗

The 'ping' tool is included in the tool suite. The ping request type was removed from the MCP spec (as of 2026-07-28). Servers should not define or advertise ping as a callable tool.

ping

HIGHconstrained-input ↗

No input validation or constraint documentation. The 'search' tool accepts a free-form query string with no maximum length, pattern, or guidance on format. The 'update' tool's content parameter is documented only as 'object' with no schema of what fields are valid or required.

searchupdate

HIGHrecovery-guide ↗

Recommendations

Formalize JSON Schema definitions for all tools. Specify input types (string, object, array), required vs. optional fields, enum constraints for choice parameters, and min/max bounds for numeric fields.
Remove the 'ping' tool or migrate its functionality. The ping request type is no longer part of the MCP spec. If health checks are needed, fold this into a capabilities or info endpoint.
Add recovery guidance to error responses. E.g., for retrieve: 'Document ID not found. Call list() to see available IDs.' For update: 'Invalid content object. Fields must match document structure.'
Implement a confirmation or dry-run pattern for the update tool. Either require explicit confirmation=true, or offer an update_preview tool that shows changes before committing.
Document output schemas for all tools. Specify the response structure, field types, and pagination metadata (total count, next_cursor if applicable). Ensure common downstream tools (e.g., update) receive IDs or references they need.
Enhance description language to be LLM-optimized. Aim for 50 - 200 characters. Include WHAT (action), WHEN (use case), and any prerequisites. E.g., 'Retrieve a document by its ID. Use this after search() to get full content. Returns title, content, and metadata.'
Add minimum/maximum bounds for query length in search. E.g., 'query: 2 - 500 characters, case-insensitive substring match'.
Clarify mutually exclusive or dependent parameters. If update accepts an optional content object, state what happens if it is omitted or partially specified.

Score history

Overall score trend

↑ 6 points across a rubric change (v1 → v2)

40/100

Scored	Grade	Overall	Spec posture	Rubric
2026-09-22	F	40	2026-07-28+	v2
2026-03-09	F	34	-	v1