A Rust SDK implementation for the Model Context Protocol, providing server and client functionality with support for multiple transport mechanisms
Scoring was not performed
Duplicate tool names ('add' appears 3 times) creates ambiguity. LLMs cannot distinguish between tools with identical names. Each should have distinct names reflecting functional difference (e.g., add_integers, add_floats, add_numbers).
No tool descriptions visible in the server definition. Per hard scoring rule: if a tool has no description, description score MUST be 0. All three 'add' tools lack non-empty descriptions required by pattern:tool-description.
Input schemas lack parameter descriptions. The visible schema shows types (integer, number) for parameters 'a' and 'b' but no descriptions explaining what these parameters control. Baseline from 549 production tools: 100% of A+ tools have parameter descriptions.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 30 | <=2025-11-25 | v2 |
| 2026-03-09 | F | 26 | - | v1 |
Schemas lack formal JSON Schema constraints (minValue, maxValue, pattern, enum). The visible schemas show only bare types with no validation guidance. This allows LLMs to pass invalid values (negative infinity, NaN for numeric tools) without constraints.
Test fixtures used instead of real server code. The three 'add' tools are defined in test files (crates/rmcp/tests/test_with_js/server.js), not production server implementations. Cannot assess real-world server quality from test stubs.
No output schema documentation. The visible tool definitions show input_schema but no documented return types or output structure. LLMs need to know what fields to expect.'
No error handling guidance in tool definitions. The test fixtures show no error recovery hints, retry guidance, or user-fixable vs fatal categorization. Per pattern:recovery-guide: 'Error responses must tell the LLM what to do next.'