MCP Server for Weighted Borda Count Preference Aggregator Skill - Pure Python Standard Library implementation of positional voting and preference aggregation. Calculates credibility-weighted Borda counts, detects Condorcet winners via pairwise matrix, and ranks multi-agent reasoning paths deterministically.
Single tool 'aggregate_votes' has a clear, descriptive name starting with a verb and a solid 180-char description explaining what it does (Borda count voting, Condorcet detection, ranked results). Input schema is present with proper JSON Schema structure and all parameters typed (candidates: array of strings, ballots: array of objects with voter_id, ranking, weight). However, parameter descriptions are minimal (e.g., 'Identifier for the voter, defaults to anon if not provided' is adequate but terse). Output schema is NOT documented, the tool returns ranking, borda_scores, and condorcet_winner but the response structure is not formally specified in the tool definition. No error handling guidance is visible (what if candidates list is empty? what if a ballot ranking contains unknown candidates?). No input validation constraints (e.g., weight bounds, ranking length limits). Tool is read-only (no state mutation), which is good, but lacks defensive documentation for edge cases.
Aggregates weighted preference ballots using Borda count voting method. Computes Borda scores for candidates, detects Condorcet winners via pairwise comparison matrix, and returns ranked results.
Output schema not documented. Tool returns {ranking, borda_scores, condorcet_winner} but LLM has no formal spec of field types, structure, or what to expect. Forces LLM to infer output shape.
No error handling guidance. Code silently filters invalid candidates from rankings (valid_ranking = [c for c in ranking if c in self.candidates]) but does not signal to LLM when data is dropped. LLM cannot know if aggregation is partial or complete.
Parameter constraints missing. 'weight' parameter has no documented bounds (negative? zero? unbounded?). 'ranking' array has no length or uniqueness constraints. LLM may pass invalid data without guidance.
Minimal parameter descriptions. 'Ordered list of candidates from highest to lowest preference' is clear, but 'Credibility weight for this ballot, defaults to 1.0' lacks guidance on valid range, interpretation, or impact on scoring.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 47 | 2026-07-28+ | v2 |
STDIO transport only. Server is not remotely accessible and cannot be used by hosted MCP clients. Hard cap at 50 for protocol readiness.