Loading ToolBench

Reading the current corpus…

langgraph-ai MCP server · F · ToolBench

langgraph-ai

A collection of Model Context Protocol (MCP) servers demonstrating various patterns including terminal command execution, web fetching, and SSE-based communication

2026-07-28+STDIO1 tools

33/100

Definition Quality26

Protocol Readiness28

Supportability49

Spec posture

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

Static source inference · medium confidence

Current-spec patterns detected

Summary

This repository is a LangGraph cookbooks/examples collection, NOT a production MCP server. The evaluation is based on the single MCP server present: fetch_server. The 'fetch' tool has a generic description with no input schema visible in the provided source. The pyproject.toml shows version 0.6.3 of 'mcp-server-fetch' but the actual tool implementation code is not included in the source dump, only requirements and Dockerfile are visible. The fetch_server is incomplete and underdeveloped for production use.

Tools (1)

fetchread only33/100

Fetch and convert web content for usage by LLMs

Issues (4)

CRITICALtool ↗

Tool implementation source code not visible, only requirements, Dockerfile, and pyproject.toml provided. Cannot verify input schema, parameter descriptions, output structure, or error handling.

fetch

HIGHtool-description ↗

Tool description 'Fetch and convert web content for usage by LLMs' is generic and lacks specificity. Does not explain WHEN to use it, WHAT formats it returns, or what prerequisites exist. Violates pattern:tool-description baseline of 194 chars avg and pattern requirement to explain WHAT, WHEN, and WHAT IT RETURNS.

fetch

CRITICALconstrained-input ↗

No input schema visible. Cannot verify parameter types, descriptions, constraints (enums, ranges), or defaults.

fetch

HIGH

Repository is primarily LangGraph educational material (cookbooks, agentic patterns examples, memory demos). Only 1 of many subdirectories is an actual MCP server (fetch_server). This is not a dedicated MCP server project suitable for production evaluation.

fetch

Recommendations

Provide complete tool implementation source code, not just dependencies and Docker configuration. The evaluator must see tool registration, input schemas, and output structure to assess quality.
Expand the 'fetch' tool description to at least 50-150 characters explaining: (1) What formats are supported (HTML, PDF, JSON)? (2) When should the LLM call this vs. a specialized scraper? (3) What does the response include (title, body, metadata, links)? (4) Are there size limits or timeouts?
Define explicit input schema with parameter descriptions. Expected parameters likely include: URL (required, description: 'The web page or API endpoint to fetch; must be http:// or https://'), format (optional, enum: ['text', 'markdown', 'json', 'html'], description: 'Output format; defaults to markdown'), timeout_seconds (optional, integer, 5-60, description: 'Request timeout in seconds; defaults to 30').
Document output schema. Likely response includes: content (string, the fetched/converted data), url (string, final resolved URL after redirects), format (string, the format returned), title (optional, string, page title or API endpoint name), timestamp (ISO 8601 string, when fetched). State any pagination support and limits (e.g. max 1 MB per fetch).
Add error handling guidance. Describe recovery for: HTTP errors (404, 403, 500), timeouts, invalid URLs, unsupported formats, size limits exceeded. E.g. 'If fetch fails with a 403 Forbidden, the resource may require authentication, consider using your own API credentials or browser session.'

Score history

Overall score trend

↑ 33 points across a rubric change (v1 → v2)

33/100

Scored	Grade	Overall	Spec posture	Rubric
2026-09-22	F	33	2026-07-28+	v2
2026-03-09	F	0	-	v1