A demonstration MCP server showcasing tools, resources, and prompts
This server implements 12 tools with basic definitions and schemas, but falls short of production quality. All tools have descriptions (10-50 chars) and input schemas with typed parameters. However, descriptions are generic and lack LLM optimization guidance; parameter descriptions are minimal (5-15 chars); output schemas are not documented; error handling lacks recovery guidance; and there is no evidence of idempotency guarantees, permission checks, or audit logging. The server is a functional demo but does not follow advanced patterns from the 54 Agentic Tool Patterns. Average per-tool score: 54/100.
Add two numbers together
Create a new note with title and content
Divide two numbers
Get a specific note by ID
Get current weather for a city
Get weather forecast for a city
List the contents of a directory
List all notes
Multiply two numbers
Tool descriptions are generic and under 50 characters; lack context for when to use each tool or how they differ from similar operations. E.g., 'Add two numbers together' provides no guidance on precision, numeric type constraints, or error handling for division by zero.
Parameter descriptions are minimal (5-15 chars). E.g., 'First operand' and 'Second operand' do not state numeric type, range, precision, or what happens if division by zero occurs. LLMs cannot infer constraints from these labels.
No output schemas documented. Tools return results but the response structure (fields, types, example data) is not declared in code or documentation. Agents cannot plan downstream calls or extract results reliably.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 38 | <=2025-11-25 | v2 |
| 2026-03-09 | F | 29 | - | v1 |
Read the contents of a file
Subtract two numbers
Write content to a file
No error handling documentation. Tools like 'divide' can fail (division by zero), 'read_file' can fail (file not found), and 'write_file' can fail (permission denied), but error recovery guidance is absent. Agents have no guidance on what to do next.
File system tools ('read_file', 'write_file', 'list_directory') expose path parameters as free-form strings with no validation description, length limits, or path traversal guards stated. No mention of sandbox boundaries or allowed directories.
'list_notes' and 'list_directory' lack pagination parameters (limit, offset, page_size) and do not state maximum result count. Returning unbounded lists risks context window exhaustion and LLM reasoning degradation.
Numeric parameters ('days' in get_weather_forecast) lack documented ranges. No constraint stated: can days be 0, 1, 100, or unlimited? Without bounds, LLMs may pass invalid values.
Tools marked as WRITE ('create_note', 'write_file') lack idempotency or confirmation patterns. No dry-run option, no idempotent key support, and agents may accidentally create duplicates or overwrite files on retry.