Quickstart examples and resources for Model Context Protocol (MCP) servers and clients across multiple languages
This is a quickstart/example repository, not a production MCP server. The codebase contains multiple incomplete tool implementations across different languages. Tools are defined in test helpers and example implementations with minimal descriptions and inconsistent schema definitions. The two primary tools (get-alerts, get-forecast) in weather-server-typescript have basic descriptions (19-22 chars each, below the 20+ baseline) and parameter descriptions that are present but minimal. Mock implementations in tests/helpers/tool-loop-test.ts define tools with even shorter descriptions. No evidence of comprehensive error handling, output schema documentation, or LLM-optimized descriptions. This reads as a teaching repository where tool quality is secondary to demonstrating MCP protocol mechanics.
Get weather alerts for a state
Get weather forecast for a location
Mock alerts, returning a top-level array
Mock forecast, returning an object
Tool descriptions are critically short (15-22 characters). Baseline for A+ tools is 50-200 chars with clear WHAT/WHEN/WHY framing. Current descriptions like 'Mock alerts, returning a top-level array' and 'Get weather alerts for a state' fail to explain when to use the tool or what fields are returned.
No documented output schemas. Tools define input parameters with JSON Schema but provide no description of return types, field names, or structure. LLMs cannot plan downstream tool calls or extract required data without knowing what fields to expect.
No error handling or recovery guidance. No indication of what can go wrong (invalid state code, unreachable API, rate limits) or what the LLM should do next. Mock implementations return dummy data; real servers would fail silently.
Inferred effective spec: 2025-06-18+.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | D | 51 | 2025-06-18+ | v2 |
| 2026-03-09 | F | 39 | - | v1 |
Parameter descriptions are minimal. 'Two-letter state code (e.g. CA, NY)' and 'Latitude of the location' lack validation constraints (latitude range -90 to 90), precision requirements, or failure cases. Descriptions do not state what happens if invalid input is provided.
Mock tools in test helpers (tools registered in tests/helpers/tool-loop-test.ts) have no separation from real implementations. Tests define identical tool names with different schemas (e.g., two 'get_alerts' and two 'get_forecast' tools), creating ambiguity about which is canonical.