Loading ToolBench

Reading the current corpus…

AgenticAI_CapstoneProject MCP server · F · ToolBench

AgenticAI_CapstoneProject

github.com/Jereshea/AgenticAI_CapstoneProject

A multi-agent e-commerce shopping assistant that finds products, compares prices, and analyzes reviews using real-time scraped data. Orchestrates remote A2A agents for price tracking, product recommendation, and review analysis.

<=2025-11-25STREAMABLE_HTTP2 tools

26/100

Definition Quality31

Protocol Readiness12

SupportabilityNot measured-

Spec posture

Inferred effective spec <=2025-11-25·current baseline 2026-07-28

Static source inference · medium confidence · detected: Logging

Deprecated protocol patterns detected

Summary

This server exposes only 2 tools with minimal schema definition and poor descriptions. Tool names lack action verbs (amazon_scraper, google_search are acceptable but generic). Descriptions are extremely brief (10-15 chars each), well below the 50-200 char baseline for LLM-optimized tools. Input schemas are present but minimal: only a single 'input_str' parameter with basic string type, no enums, no validation constraints, no output schema documentation. No error handling guidance, no parameter dependencies documented, no pagination support despite tools returning lists. The server is built on A2A (Google's Agent API), not MCP, it does not register tools via MCP protocol at all. Tool definitions are inferred from agent skill metadata, not explicit MCP tool registration.

Tools (2)

amazon_scraperread onlysource verified30/100

Fetch price of the first Amazon product.

google_searchread only27/100

Search the internet using Google Search to find product recommendations and reviews.

Issues (5)

CRITICALtool-description ↗

Tool descriptions are critically short (10-15 chars). 'Fetch price of the first Amazon product' and 'Search the internet using Google Search...' lack context for LLM selection. Baseline is 50-200 chars; these are 1/5 the minimum.

amazon_scrapergoogle_search

CRITICALconstrained-input ↗

Input schemas are minimal. Both tools accept only 'input_str' (string type). No enums, no format constraints, no validation rules, no min/max length. LLMs cannot infer valid input ranges or formats.

amazon_scrapergoogle_search

CRITICALtool ↗

No output schemas documented. LLMs do not know what fields to expect from amazon_scraper or google_search results. Cannot plan downstream tool calls or extract structured data.

amazon_scrapergoogle_search

CRITICALtool ↗

Server is built on A2A (Google Agent API), not MCP. Tools are registered as AgentSkill objects in A2A, not via MCP protocol. This is a fundamental architecture mismatch, the server cannot be used as an MCP server by MCP clients.

Recommendations

Expand tool descriptions to 50-200 chars. Example: 'amazon_scraper: Fetch the current price, availability, and rating of the first matching product from Amazon. Use this to compare prices across products or check if an item is in stock. Returns price (USD), availability status, and product URL.'
Add input parameter constraints. Replace bare 'input_str' with: type='string', minLength=3, maxLength=100, description='Product name or ASIN to search (e.g., "Samsung Galaxy S24"). Partial names are OK; the tool returns the first match.'
Document output schemas. Specify what amazon_scraper returns: {price: number, currency: string, availability: enum[in_stock|out_of_stock|preorder], product_url: string, rating: number}. Same for google_search.
Add error handling hints. E.g., 'If no product found, try a shorter or more generic name. If timeout occurs, retry once after 2 seconds.'
Migrate from A2A to MCP protocol. Register tools via MCP's tool registration mechanism (CallToolRequest/CallToolResult) instead of A2A AgentSkill. This enables compatibility with MCP clients and hosted platforms.

Spec posture evidence

Inferred effective spec: <=2025-11-25.

Relies on Logging (deprecated) - log to stderr or use OpenTelemetry

Score history

Overall score trend

First recorded score · v2 rubric

26/100

Scored	Grade	Overall	Spec posture	Rubric
2026-09-23	F	26	<=2025-11-25	v2