MCP Server implementing Wang et al. Self-Consistency majority voting for clustering and consensus from multiple candidate responses
Single tool with minimal schema and description. Tool description is 67 chars, acceptable but generic. No output schema documented. No error handling guidance. No parameter constraints or validation hints. Tool name 'calculate_majority_vote' is verb-noun compliant but the implementation lacks production-grade rigor expected for agent integration.
Calculate consensus majority vote from candidate responses.
Input schema missing parameter descriptions. 'candidates' array has no description explaining what strings it accepts, expected format, or constraints.
No output schema documented. LLM cannot predict response structure (consensus_answer, confidence, distribution, etc.) before calling the tool.
No error handling or recovery guidance. Tool does not document what happens on empty input, malformed candidates, or edge cases. LLM has no guidance on next steps if call fails.
Tool description lacks context on when to use it. Does not explain prerequisites (e.g., 'requires multiple candidate responses'), expected input format, or what 'majority vote' means in this context.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 42 | 2026-07-28+ | v2 |
No input validation hints. Description does not specify minimum/maximum candidates, expected string length, or handling of duplicates or empty strings.