MCP Elasticsearch Server with Semantic S2L Layer. Provides tools for querying Elasticsearch with semantic field descriptions and business rules externalized in JSON configuration.
Server has 7 tools with clear naming (verb_noun pattern: cluster_info, list_indices, get_mappings, get_field_descriptions, search, esql, get_shards). All tools have descriptions (10-200 chars, within baseline 34-392 range). Input schemas are present with type definitions and parameter descriptions. However, output schemas are NOT documented, tool descriptions state what they return (e.g., 'Return Elasticsearch cluster information') but the actual response structure is not formally specified. Error handling is minimal: no recovery guidance, no error categorization, no actionable error messages. Parameters lack constraints (enums, ranges, patterns), e.g., 'index' accepts any string, 'size' has no bounds. The 'search' tool accepts complex nested objects (query, aggs) with minimal guidance on structure. No tool declares permissions or security implications. Composition is sound (each tool does one thing), but chaining is not optimized, e.g., list_indices returns index names, but get_mappings requires an index name, forcing sequential calls. Overall: solid naming and basic schemas, but missing output documentation, error guidance, and input validation.
Return Elasticsearch cluster information. Includes version and available features (e.g. ES|QL availability).
Execute an ES|QL query (Elasticsearch SQL-like language). ES|QL is only available from Elasticsearch 8.11.0 onwards.
Return field descriptions and business rules from the S2L semantic layer. IMPORTANT: Call this tool WITHOUT arguments before any business query to get all business rules (mandatory filters, billing criteria, error codes, timezone handling, index patterns, etc.). Only use field_pattern if you are looking for a specific known field.
Return the mapping (schema) of an Elasticsearch index, enriched with field descriptions and business rules from the S2L semantic layer.
Return shard information for an index.
Output schemas not documented. Tool descriptions state what they return (e.g., 'Return cluster information') but the actual response structure (fields, types, nesting) is not formally specified. LLMs cannot plan downstream tool calls or extract fields without knowing the response schema.
No input validation or constraints. 'size' parameter in search() has no min/max bounds (could accept 0 or 1000000). 'index' accepts any string without pattern validation. 'query' in search() is a free-form object with no schema guidance. Unbounded inputs invite hallucinated values and API errors.
Error handling is absent. No recovery guidance, error categorization, or actionable messages. If a search fails or an index is not found, the LLM receives a raw exception with no guidance on what to do next (retry? call a different tool? ask the user?).
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 49 | 2026-07-28+ | v2 |
List available Elasticsearch indices.
Execute a Query DSL search against Elasticsearch.
No pagination support. 'list_indices' and 'search' may return large result sets, but neither accepts limit/offset parameters or returns a total count. Large responses blow context windows and degrade LLM reasoning.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). All tools are read-only, but this is not declared in the tool definitions. Agents cannot distinguish safe tools from destructive ones without explicit hints.