MCP server providing cryptocurrency news, analytics, social sentiment, price data, and token information from the CryptoMinute API
The server registers 8 tools with descriptions and claims to load schemas from embedded JSON files. However, the actual schema files are not provided in the source code, so schema completeness cannot be verified. Tool names use action-adjacent patterns (search, monitor, insights) but lack consistent verb-first naming. Descriptions are present but quite brief (avg ~40 chars), and critically, no parameter descriptions are visible in the source, only schema file paths are referenced. The code shows s.loadSchema() loading from 'schemas/<toolName>.json', but these files are not included in the provided source, making schema validation impossible. Per the hard scoring rules, tools with unverifiable schemas cannot score above 50 on schema dimension.
Five-dimensional narrative analytics
Real-time flash posts from cryptocurrency Telegram channels
Filter news by ticker, sentiment, date, significance
Historical OHLCV data
Reddit sentiment & impact
AI-clustered narrative summaries
Token metadata & categorization
Schema files not provided in source code. The code references 'schemas/*.json' via embed.FS, but these files are absent from the provided codebase. Cannot verify parameter types, required fields, or nested object structures.
Parameter descriptions missing. The code references schema files but they are not provided. Without visible parameter descriptions, cannot assess whether LLM can understand each parameter's purpose, valid values, constraints, or format. Patterns require 'Add a non-empty description to every input parameter.'
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 40 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 35 | 2024-11-05+ | v1 |
Video engagement metrics
Tool descriptions are too brief (30-45 chars). Descriptions like 'Historical OHLCV data' and 'Token metadata & categorization' lack context about WHEN to use the tool, WHAT it returns, and any prerequisites. Baseline indicates descriptions should average 194 chars for clarity. Current descriptions provide minimal LLM guidance for tool selection.
Naming inconsistency: tools lack consistent verb-first pattern. 'crypto-news-search' uses search verb, but 'crypto-analytics', 'crypto-stories', 'crypto-tokens', and 'crypto-flash-posts' lack action verbs. Should be: 'search-crypto-analytics', 'list-crypto-stories', 'list-crypto-tokens', 'search-crypto-flash-posts' etc. Baseline: 90% of A+ tools start with action verbs (get, list, create, search, update).
No output schema documentation. Code does not document what fields tools return or their types. LLMs cannot plan multi-step workflows without knowing what data is available for downstream tool calls. Pattern requires: 'Document the output schema. LLMs need to know what fields to expect.'
Parameter type annotations missing from visible code. JSON provided in the evaluation shows parameter types in the InputSchema (e.g., 'type': 'string', 'type': 'number'), but these are not verified in the actual schema files since they are not provided. Cannot confirm required fields, min/max constraints, enums, or format patterns.
Enum parameters should be enforced where applicable. Parameters like 'sort', 'order', 'status', and 'scopes' likely accept limited values but no enum validation is visible. Free-form strings invite hallucinated values. Pattern: 'When a parameter accepts one of a known set of values, declare it as an enum.'
No error handling guidance in visible code. The MCPError struct is defined but no evidence of actionable error messages, recovery suggestions, or error classification (retryable vs user-fixable). Pattern requires: 'Error responses must tell the LLM what to do next.'
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). All tools are marked Risk: READ_ONLY but code does not emit tool annotations in the tool registration. Spec 2026-07-28 rewards tool annotations for clarity.