exasol-mcp-server

Exasol MCP Server

Optional Tools

Several tools — including execute_exasol_query, profile_exasol_query, execute_exasol_write_query, summarize_exasol_table, and all BucketFS tools — are disabled by default and may not appear in the tool list. If a workflow below requires a tool that is not available, do not retry or attempt workarounds: inform the user that the tool is disabled and needs to be enabled by the server administrator.

Common Workflows

Schema Exploration

Start broad, then narrow down:

1. list_exasol_schemas
   → pick a schema

2. list_exasol_tables_and_views(schema_name=...)
   → pick a table

3. describe_exasol_tables_and_views(schema_name=..., table_names=[...])
   → see columns, types, constraints (pass every table/view needed in one call)

4. summarize_exasol_table(schema_name=..., table_name=...)
   → understand data distribution (if enabled)

If you already know roughly what you are looking for, use the find_* variants to keyword-search across all schemas at once.

Writing and Debugging a Query

1. execute_exasol_query(query="SELECT ...")
   → verify the query works and returns expected data

2. profile_exasol_query(query="SELECT ...")
   → check the execution plan if the query is slow
   → look at the EXECUTION_MODE column: COMPILE, EXECUTE, WAIT
   → high COMPILE time → query rewrite or simplification may help
   → high EXECUTE time → check JOIN strategies and data distribution

Discovering and Using UDFs

1. list_exasol_user_defined_functions(schema_name=...)
   → see all UDF scripts in a schema

2. describe_exasol_user_defined_functions(schema_name=..., func_names=[...])
   → read source code and parameter declarations (pass every UDF needed in one call)

3. execute_exasol_query(query="SELECT schema.script_name(col) FROM table")
   → invoke the script on data

BucketFS Workflow

1. list_bucketfs_directories(path="/")
   → see top-level buckets

2. list_bucketfs_files(path="/my_bucket/models/")
   → list model files

3. read_bucketfs_text_file(path="/my_bucket/config.json")
   → read a configuration file

4. write_text_to_bucketfs_file(path="/my_bucket/output.txt", content="...")
   → save results or configuration