MCP server providing tools to query and retrieve metadata from Oracle Eloqua databases via JDBC
The server defines 3 data-access tools with clear verb-based naming and reasonable structure. However, there are significant gaps: output schemas are entirely undocumented (CSV format mentioned in descriptions but no formal schema definition), parameter descriptions lack specificity about constraints and formats, and error handling is minimal. The tools follow basic registration patterns but fall short of production-grade quality in schema completeness and LLM guidance.
Retrieves a list of fields, dimensions, or measures (as columns) for an object, entity or collection (table). Use the `eloqua_get_tables` tool to get a list of available tables. The output of the tool will be returned in CSV format, with the first line containing column headers.
Retrieves a list of objects, entities, collections, etc. (as tables) available in the data source. Use the `eloqua_get_columns` tool to list available columns on a table. Both `catalog` and `schema` are optional parameters. The output of the tool will be returned in CSV format, with the first line containing column headers.
Execute a SQL SELECT statement.
Output schemas completely undocumented. All three tools return CSV format but no formal response schema is defined. LLMs cannot plan downstream processing or extract specific fields without guessing the structure.
Parameter descriptions lack format and constraint details. 'catalog', 'schema', and 'table' parameters have minimal descriptions that don't explain valid formats, whether they're case-sensitive, required vs optional nuances, or character restrictions. SQL parameter in eloqua_run_query mentions 'mostly based around SQL-92' and allowed clauses but doesn't clarify the exact dialect quirks or whether subqueries are supported.
Error handling is generic and non-recoverable. Code shows `throw new RuntimeException('ERROR: ' + ex.getMessage())` with no guidance on what the LLM should do next. Errors are not categorized (retryable vs fatal). No hint tools to resolve failures (e.g., 'table not found, call eloqua_get_tables to discover available tables').
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 44 | <=2025-11-25 | v2 |
| 2026-03-09 | D | 50 | - | v1 |
Tool descriptions are moderately complete but lack actionable context. Descriptions explain what each tool does but don't explain WHEN to use it or what happens with edge cases. For example, eloqua_get_tables says it returns tables but doesn't explain if it will error on invalid catalog/schema or return empty results.
No input validation or cleansing visible in code. Parameters are passed directly to JDBC calls without sanitization, risking SQL injection if the LLM is tricked into crafting malicious SQL via the 'sql' parameter in eloqua_run_query.
eloqua_run_query description is vague about SQL dialect support. Says 'mostly based around SQL-92' and lists allowed clauses but doesn't specify what's NOT supported (CTEs, window functions, etc.) or what vendor-specific extensions Eloqua adds. LLMs may attempt unsupported syntax.