MCP server for WooCommerce database access via JDBC, providing tools to list tables, retrieve columns, and execute SQL queries
The server implements three well-structured tools with clear naming conventions and appropriate descriptions. Tool names follow verb_noun patterns (get_tables, get_columns, run_query). However, there are significant gaps in schema completeness, parameter descriptions, and error handling guidance. The tools accept parameters but many lack type constraints and detailed descriptions. Output format is mentioned (CSV) but not formally documented. Error handling throws generic RuntimeException without actionable recovery guidance. The schema builder shows proper construction, but underlying validation and error classification are absent. Overall this is a fair implementation with noticeable gaps typical of community servers.
Retrieves a list of fields, dimensions, or measures (as columns) for an object, entity or collection (table). Use the `{prefix}_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 `{prefix}_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.
Missing output schema documentation. Tools return CSV format but provide no formal schema definition of the returned data structure. LLMs cannot plan downstream operations or validate results without knowing field names and types.
run_query tool description lacks critical constraints and safety guidance. No mention of query timeout, result size limit, supported clauses beyond basic examples, or SQL injection prevention. Description mentions 'mostly based around SQL-92' without specifying unsupported clauses, inviting invalid queries.
Parameter descriptions lack detail about constraints and formats. Catalog and schema parameters are described as 'The catalog name' and 'The schema name', minimal guidance on whether they are optional, what values are valid, or how to discover available values. Table parameter in get_columns is marked required but description doesn't explain how to find table names.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | D | 50 | <=2025-11-25 | v2 |
| 2026-03-09 | F | 36 | - | v1 |
Error handling is non-existent. All exceptions are caught and rethrown as generic RuntimeException('ERROR: ' + message). No actionable recovery guidance, error categorization, or next-step hints. LLMs cannot determine if errors are retryable, user-fixable, or fatal.
Input validation is missing. The run_query tool accepts arbitrary SQL without validation of dangerous patterns (DROP, DELETE, INSERT, UPDATE). While marked READ_ONLY, the tool provides no server-side enforcement. Schema parameters accept null/empty strings without validation.
No pagination support documented. Tools that return lists (get_tables, get_columns) provide no limit, offset, or next_cursor parameters. If a WooCommerce instance has thousands of products or columns, returning all results will exceed token budgets and degrade LLM reasoning.
Tool descriptions do not explain when to call each tool or what dependencies exist. A user asking 'list all products' may not know to call get_tables first, then get_columns, then run_query. No 'discovery' guidance or multi-step planning hints.
Parameter naming lacks type suffixes. Parameters named 'catalog', 'schema', 'table' are ambiguous, they could be IDs, names, or paths. No suffix like 'table_name' to disambiguate what format the LLM should pass. For run_query, the 'sql' parameter is clear but lacks constraints like max length or max result rows.
No idempotency or confirmation mechanism for run_query. SQL SELECT is read-only, but a complex query could trigger expensive computations or lock tables. No dry-run option or explicit timeout communication to the LLM.