MCP App Store
by KeenEthics
Cognitive Database icon

Cognitive Database MCP Server for ChatGPT

By Transwarp Technology (Singapore) Pte. Ltd.

Overview

Bring the power of Cognitive Database into your conversations. Explore databases, execute GPU-accelerated SQL queries, analyze data, and uncover insights through natural conversation. Connect directly to your cloud data environment to discover accessible tables, investigate trends and anomalies, assess data quality, and turn query results into clear explanations. Work with your data from preparation to analysis without switching between tools. Load CSV and Parquet files into tables, upload SQL files, publish supported documents to knowledge volumes, connect external data sources, and manage ingestion workflows. Organize data assets, manage Compute resources and access, and inspect usage and billing to understand how your database resources are used. Ask questions in your own words, explore the data behind an answer, and move from an initial question to a query, an explanation, and a practical next step.

How to connect

  1. ChatGPT: Open the Cognitive Database listing in the ChatGPT app directory and choose Connect. ChatGPT listing Cognitive Database exposes 134 tools here.

Tools

cognitivedb_audit_logs_list

ChatGPT
List audit events scoped to a Catalog, Schema, or Table using an explicit resource_type and exact resource-name hierarchy. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_catalog_create

ChatGPT
Create a managed catalog in the current space. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_catalog_delete

ChatGPT
Delete a Catalog selected by its exact name. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_catalog_delete_impact_check

ChatGPT
Preview what would be affected by deleting a Catalog, including contained schemas and tables, direct grants, affected grantees, and current-Space visibility. Always check before deletion. Does not delete the Catalog. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_catalog_list

ChatGPT
List catalogs in the current space through Sophon Mind.

cognitivedb_catalog_permissions_get

ChatGPT
Get the current user's effective capabilities for a Catalog, including preview, edit, delete, ownership transfer, and permission management. Only returns capabilities for the current user.

cognitivedb_catalog_permissions_grant

ChatGPT
Grant selected Catalog privileges to a grantee. Check with the user that the Catalog, grantee, and selected privileges are correct before executing. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_catalog_permissions_list

ChatGPT
List direct and inherited permission grants for an exact Catalog, optionally narrowed to a Schema or Table name hierarchy.

cognitivedb_catalog_permissions_revoke

ChatGPT
Revoke selected direct Catalog privileges from a grantee. Check with the user that the Catalog, grantee, and selected privileges are correct before executing. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_catalog_schemas

ChatGPT
List schemas for up to 5 Catalogs selected by exact name. The Tool prechecks all Catalog names, queries every valid Catalog, and reports per-Catalog errors without discarding successful results.

cognitivedb_catalog_table_columns

ChatGPT
List columns for up to 5 Tables grouped by Catalog and Schema in targets. The Tool prechecks every exact Table name, resolves Studio coordinates once per Catalog Schema, and reports per-Table errors without discarding successful results.

cognitivedb_catalog_tables

ChatGPT
List tables for up to 5 Schemas grouped by Catalog in targets. When a Schema returns more than 100 tables, its items contain only name. Check each result's truncated flag and message for completeness and next steps. A failed Schema does not discard other results.

cognitivedb_catalog_update

ChatGPT
Replace the description of a Catalog. Check with the user that the Catalog and final description are correct before executing. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_cloud_ingestion_file_browse

ChatGPT
Browse folders and files in one S3 or Azure Blob connection for Cloud Ingestion. Pass the exact connection name and connectionType from the same CONNECTED result returned by cognitivedb_cloud_storage_connection_list. Use a folder or Parquet file path from the result with cognitivedb_cloud_ingestion_file_preview, then carry the same connection and selected path into cognitivedb_cloud_ingestion_task_create.

cognitivedb_cloud_ingestion_file_preview

ChatGPT
Preview the columns and sample rows of an S3 or Azure Blob Parquet file or directory selected through cognitivedb_cloud_ingestion_file_browse. Pass the exact connection, connectionType, and sourcePath returned by the browse chain. Returns the sampled file, resolved path, detected columns, sample rows, and suggested target table name.

cognitivedb_cloud_ingestion_task_create

ChatGPT
Validate and create a manual full-load Cloud Ingestion task for an S3 or Azure Blob Parquet source selected through cognitivedb_cloud_ingestion_file_preview. Pass the exact connection, connectionType, and sourcePath from the same browse/preview chain. Validation warnings block creation; after the user accepts validation.warnings, repeat the same request with acknowledgedWarnings set to that complete warning list. The exact source and target are validated before creation. Pass the IANA timezone to save with the task.

cognitivedb_cloud_storage_connection_check

ChatGPT
Test one saved S3 or Azure Blob connection by its exact Connection name and connectionType. Connector uses saved credentials and updates the saved connection status and timestamp. Returns a stable testedConnection summary without the pre-test status, the normalized test result without free-form upstream messages, and credential-free provider-specific repair guidance. It does not accept or return Role ARN, External ID, Access Key, Account Key, Azure tenant/client identifiers, Client Secret, or other credential values.

cognitivedb_cloud_storage_connection_delete

ChatGPT
Delete one saved S3 or Azure Blob connection. First use cognitivedb_cloud_storage_connection_list or cognitivedb_cloud_storage_connection_show, then pass the exact current Connection name and matching expectedConnectionType from the same result. Deleting the saved connection does not delete its bucket, container, stored objects, or AWS/Azure identity resources, but dependent workflows may stop working. Do not retry automatically when the deletion outcome is unknown; use show or list to determine whether it still exists.

cognitivedb_cloud_storage_connection_list

ChatGPT
List configured cloud storage connections with provider details, authentication method, status, and scope. Omit connectionType to list all currently supported types, or specify S3 or AZURE_BLOB to filter. Azure Blob authMethod is ACCOUNT_KEY or SERVICE_PRINCIPAL. Results are paginated; continue while hasMore is true. Pass the exact connection name and connectionType from the same CONNECTED S3 or Azure Blob result through cognitivedb_cloud_ingestion_file_browse and cognitivedb_cloud_ingestion_file_preview; then pass the selected connection name, connectionType, and sourcePath to cognitivedb_cloud_ingestion_task_create.

cognitivedb_cloud_storage_connection_show

ChatGPT
Get the current public details of one configured S3 or Azure Blob connection by its exact Connection name and connectionType. Returns its provider type authentication method, status, scope, and provider details using the same credential-free structure as cognitivedb_cloud_storage_connection_list. Use this tool to inspect a known connection without manually traversing list pages or to verify whether it still exists after an earlier operation had an unknown outcome.

cognitivedb_compute_clear_cache

ChatGPT
Clear retained cache for one stopped Compute when dataCache.canClear=true. Preserves business data; initial queries may be slower. Observe CLEARING instead of resubmitting. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_compute_create

ChatGPT
Create a compute using a compute type and available specCode from the discovery tools. category is required: choose 'default' for SQL and analytics or 'general' for notebook and interactive workloads. Call cognitivedb_compute_type_list with the same category first. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_compute_delete

ChatGPT
Delete a default or general compute. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_compute_describe

ChatGPT
Describe a compute.

cognitivedb_compute_endpoint_details

ChatGPT
Get PostgreSQL connection details for a compute and Catalog, including endpoint, database, SSL mode, and psql/JDBC connection examples. Credentials not included.

cognitivedb_compute_global_configuration_get

ChatGPT
Get global Compute configuration defaults and allowed ranges, including auto-suspend timing.

cognitivedb_compute_list

ChatGPT
List computes in the current space. category is required; use 'all' when the user did not request a specific workload category. Results identify each compute's category and supported operations. Use cognitivedb_compute_describe for one compute's full details.

cognitivedb_compute_monitoring_list

ChatGPT
List or summarize the query monitoring snapshot for the current activation of a started compute. Use it to inspect current query activity and compute utilization. For usage across activation periods, refer to query history.

cognitivedb_compute_set_default

ChatGPT
Set a default-category data-warehouse compute as the space default. General computes do not support this operation. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_compute_spec_list

ChatGPT
List available compute specification families for a compute type.

cognitivedb_compute_start

ChatGPT
Start a default or general compute. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_compute_stop

ChatGPT
Stop a default or general compute. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_compute_suspend

ChatGPT
Suspend a default-category data-warehouse compute. General computes do not support suspend; use stop instead. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_compute_type_list

ChatGPT
List compute products for a required workload category. 'default' returns data-warehouse products; 'general' returns notebook/interactive products.

cognitivedb_compute_update

ChatGPT
Update a default-category data-warehouse compute. General computes do not support update. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_database_connection_check

ChatGPT
Run a live connectivity and metadata-access test for one saved Databricks, Snowflake, PostgreSQL, or MySQL database connection by its exact Connection name. Callers must not supply credentials, a JDBC URL, Driver ID, or Provider type. The check may contact the external database, but it does not modify the saved connection or start metadata collection. Returns only a credential-free result summary and never returns saved credentials or raw connection configuration. Use cognitivedb_database_connection_show for a saved-state overview without a live test.

cognitivedb_database_connection_delete

ChatGPT
Delete one saved Databricks, Snowflake, PostgreSQL, or MySQL database connection. First use cognitivedb_database_connection_list or cognitivedb_database_connection_show, then pass the exact current Connection name and matching expectedSourceType from the same current result. This deletes the saved connection and may stop dependent ingestion workflows; it does not modify the external database or delete external database objects. Do not retry automatically when the deletion outcome is unknown; use show or list to determine whether the connection still exists.

cognitivedb_database_connection_list

ChatGPT
List saved Databricks, Snowflake, PostgreSQL, and MySQL database connections in the current workspace for Database Ingestion. Optionally filter by sourceType; omit it to query all supported types. Returns paginated, credential-free connection status and ingestion readiness derived from the saved connector status; it does not run a live connection test. Continue while hasMore is true. Use cognitivedb_cloud_storage_connection_list instead for S3 or Azure Blob file connections.

cognitivedb_database_connection_metadata_collect

ChatGPT
Start one asynchronous metadata collection batch for a saved Databricks, Snowflake, PostgreSQL, or MySQL database Connection. First use cognitivedb_database_connection_list or show, then pass the exact current Connection name and expectedSourceType from the same recent result. The Tool reuses the saved authentication and Metadata Scope, may contact the external database, and updates the saved metadata snapshot; it does not change Connection settings or return metadata immediately. A successful call only means the asynchronous request was accepted. Do not retry automatically when the outcome is unknown or metadata is still being collected. The Tool never returns JDBC URLs, Driver IDs, identities, credentials, or raw connection configuration.

cognitivedb_database_connection_show

ChatGPT
Show a saved Databricks, Snowflake, PostgreSQL, or MySQL database connection in the current workspace by its exact Connection name. Returns its saved status, Database Ingestion readiness, authentication method, credential presence, metadata scope, configured metadataSchedule, normalized metadata collection status, and last collection time. When includeSettings is false or omitted, the saved-state overview excludes the saved JDBC URL, Username and Client ID. With includeSettings=true, it also returns the readable saved JDBC URL, Username, Client ID and metadata selections; URLs containing embedded credentials are null. Passwords, Client Secrets, private key material, Driver ID, KMS Key ID and raw connection configuration are never returned. This does not run a live connection test or perform a save, metadata collection or ingestion. Use cognitivedb_cloud_storage_connection_show instead for S3 or Azure Blob connections.

cognitivedb_database_ingestion_source_browse

ChatGPT
Browse Catalogs, Schemas, or Tables in a connected Databricks, Snowflake, or PostgreSQL database connection for Database Ingestion. Set connection from cognitivedb_database_connection_list. For CATALOG, omit catalog and schema. For SCHEMA, provide the exact Catalog name. For TABLE, provide the exact Catalog and Schema names. The database provider is resolved automatically from the Connection name. Returns paginated metadata nodes, not columns or sample rows. Continue while hasMore is true; use the selected names for table-column inspection and task creation.

cognitivedb_database_ingestion_table_columns_list

ChatGPT
List columns for a Database Ingestion source Table. Set tableId to the exact TABLE nodeId returned by cognitivedb_database_ingestion_source_browse. Optionally pass the corresponding tableName to make the result and not-found error easier to read; tableId remains the lookup and existence-check key. Returns paginated column names, source types, nullability, primary-key flags, comments, and stable zero-based positions; it does not return sample rows or trigger metadata synchronization. Continue while hasMore is true. Use these columns to choose selected fields or an incremental identifier for the next Database Ingestion task.

cognitivedb_database_ingestion_task_create

ChatGPT
Create an OFFLINE, manual, single-table Database Ingestion task from one Databricks, Snowflake, or PostgreSQL source Table into a new Cognitive Database target Table. Pass the exact Connection name and source Catalog, Schema, and Table names from the preceding Database Ingestion discovery chain. The Tool resolves that hierarchy and current column metadata, derives all source names and types, checks that the target Table does not already exist, and then creates the CONFIGURATION task. Omit selectedColumns to import every current source column. INCREMENT_APPEND requires a selected incrementIdentifierColumn whose source type is eligible for that database provider. This operation only saves an OFFLINE task definition; it does not publish or run ingestion. Pass the IANA timezone to save with the task, then pass the returned taskId to cognitivedb_ingestion_task_publish next.

cognitivedb_external_metadata_list

ChatGPT
List one page of Catalog, Schema, or Table metadata from a connected Database Connection. For CATALOG, omit catalog and schema. For SCHEMA, set catalog; for TABLE, set catalog and schema. searchText filters names at the selected level by case-insensitive substring. Continue pages while hasMore is true. TABLE items include nodeId but no columns or data rows. Pass the exact Connection, Catalog, and Schema names to cognitivedb_external_table_columns_list, with selected Table names in tables even for one Table. If the Catalog or Schema is unknown, find the Table with cognitivedb_external_metadata_search first.

cognitivedb_external_sql_execute

ChatGPT
Execute one SELECT, DDL or DML statement through Studio transporter.execute_sql. Supports external connector datasources in the current Space. Cognitive Database execution must use cognitivedb_sql_query_execute, cognitivedb_sql_dml_execute or cognitivedb_sql_ddl_execute. Returns result rows synchronously; non-query statements return an empty array, not an affected-row count. All external statements, including SELECT, use standard MCP confirmation when enabled: functions and views can have side effects. No automatic retry after an uncertain outcome.

cognitivedb_external_table_columns_list

ChatGPT
List source Table columns by exact Connection, Catalog, Schema, and Table names. Pass tables as 1 to 20 unique Table names in the same Schema, including a one-element array for one Table. Results always contain tables[], each with its tableName, tableId, columns, and hasMore. page and pageSize apply to every Table; increment page and include only Tables with hasMore=true. pageSize defaults to 50 per Table, with at most 1000 returned columns and four concurrent reads. A failed Table read fails the batch. Backend column payloads are fetched before MCP pagination. No source rows are read and metadata is not refreshed.

cognitivedb_ingestion_execution_cancel

ChatGPT
Request cancellation of one active Cloud or Database Ingestion execution. Pass an exact executionId returned by cognitivedb_ingestion_task_execution_list, cognitivedb_ingestion_task_execute, or cognitivedb_ingestion_execution_show. The Tool first reads Run Detail, infers the ingestion type from FILE_INGESTION or CONFIGURATION mode, and rejects DAG executions. It submits cancellation only when Studio status is PENDING or RUNNING; an already CANCELLED execution is returned unchanged, and every other status is rejected. A successful response means cancellation was requested, not that execution has reached CANCELLED; call cognitivedb_ingestion_execution_show to verify the final state. This is a state-changing operation. Do not retry automatically when the outcome is unknown; inspect the same execution first because cancellation may already have been accepted.

cognitivedb_ingestion_execution_rerun

ChatGPT
Create a new execution from one terminal Cloud or Database Ingestion execution. Pass the exact taskId and executionId returned together by task execution list or show Tools. The Tool reads Run and Task Detail, verifies their relationship, requires the Task to be ONLINE, infers the ingestion type, and rejects DAG. Source Studio status must be COMPLETED, FINISHED, FAILED, RECOVERABLE, or CANCELLED; active, PAUSED, and UNKNOWN executions are rejected. For Cloud Ingestion, omit resultIds and the complete execution is rerun. For Database Ingestion, resultIds is required: select one or more exact numeric results[].resultId values returned by cognitivedb_ingestion_execution_show; the Tool verifies ownership and reruns only those table results. This operation creates a new execution and may write target data again. It is not idempotent. Do not retry automatically when the outcome is unknown; call cognitivedb_ingestion_task_execution_list for the same task and inspect recent executions first.

cognitivedb_ingestion_execution_result_logs

ChatGPT
List logs for one table-level result of a Cloud or Database Ingestion execution. Pass the exact executionId and resultId returned together by cognitivedb_ingestion_execution_show. The Tool reads Run Detail to infer the ingestion type and reject DAG executions, then verifies that resultId belongs to executionId before reading logs. Studio returns the complete log list, so level, page, and pageSize are applied by MCP while preserving Studio order. Each log content value is redacted while preserving line structure, then limited to 2048 characters by keeping its beginning and end; inspect contentTruncated, originalLength, and omittedCharacters when present. Logs from a non-final execution may still grow, which is indicated by logsComplete=false. This Tool is read-only and does not poll, execute, cancel, or rerun anything.

cognitivedb_ingestion_execution_show

ChatGPT
Show the current snapshot and table-level results of one Cloud or Database Ingestion execution. Pass an exact executionId from cognitivedb_ingestion_task_execution_list, cognitivedb_ingestion_task_execute, or a future rerun result. Studio addresses executions directly by executionId, so taskId and ingestionType are not inputs; the Tool reads Run Detail, infers the ingestion type from FILE_INGESTION or CONFIGURATION mode, and rejects DAG executions before reading results. It returns immediately without polling, sleeping, cancelling, or rerunning the execution. Results can be empty or incomplete while terminal is false. Use cognitivedb_ingestion_execution_wait for PENDING, DISPATCHING, or RUNNING; PAUSED and UNKNOWN require attention and should be inspected before calling Show again. This Tool is read-only and omits parameters, logs, credentials, and unknown fields. Pass one returned resultId with the same executionId to cognitivedb_ingestion_execution_result_logs when table-level logs are needed.

cognitivedb_ingestion_execution_wait

ChatGPT
Wait for a Cloud or Database Ingestion execution returned by cognitivedb_ingestion_task_execute. The Tool reads Run Detail and infers the ingestion type from FILE_INGESTION or CONFIGURATION mode, so callers do not provide a task or ingestion type. It polls until a normalized terminal status is available, then returns the run summary and table results. Timing out does not cancel the execution; call this Tool again with the same executionId to continue waiting.

cognitivedb_ingestion_task_delete

ChatGPT
Permanently delete one OFFLINE Cloud or Database Ingestion task definition. Pass the exact taskId and matching expectedName from the same cognitivedb_ingestion_task_list or cognitivedb_ingestion_task_show result. The Tool reads Task Detail, infers the ingestion type, rejects DAG and ONLINE tasks, and blocks deletion while any execution is active or unresolved. It rechecks the task identity and OFFLINE status immediately before DELETE. This does not delete source or target data or cancel executions; existing execution records and results are retained. Do not retry automatically when the deletion outcome is unknown; call task show or list first.

cognitivedb_ingestion_task_execute

ChatGPT
Submit one manual execution for an ONLINE Cloud or Database Ingestion task. Pass the exact taskId returned by cognitivedb_ingestion_task_publish; the Tool reads Task Detail and accepts only FILE_INGESTION (Cloud) or CONFIGURATION (Database), so callers do not provide an ingestion type. For Database Ingestion, the saved runtime configuration is checked before submission. The Tool will return an executionId. Do not retry automatically when the submission outcome is unknown because another run may already have started.

cognitivedb_ingestion_task_execution_list

ChatGPT
List executions belonging to one Cloud or Database Ingestion task. Pass an exact taskId from cognitivedb_ingestion_task_list, cognitivedb_ingestion_task_show, or a task-create result. The Tool reads Task Detail, infers the ingestion type, and rejects DAG tasks before reading run history. Studio returns the complete run history for a task, so page and pageSize apply MCP-local pagination while preserving Studio order. Returns credential-free execution summaries without table results, logs, task configuration, or workspace identifiers. Continue while hasMore is true. Pass an executionId to cognitivedb_ingestion_execution_show for its immediate Detail and current Results; pass an active executionId to cognitivedb_ingestion_execution_wait when you need to wait for its terminal result. This Tool is read-only.

cognitivedb_ingestion_task_list

ChatGPT
List Cloud and Database Ingestion tasks in the current workspace without returning full source, mapping, target, or execution details. Optionally filter by ingestionType, publishStatus, or task name text; omit ingestionType to include both supported ingestion types. DAG tasks are always excluded. Returns credential-free task summaries, accurate ingestion-type totals, and pagination metadata. Continue while hasMore is true only if more task details are needed; total and ingestionTypeTotals already cover all matching tasks, not just this page. This Tool is read-only and does not publish or execute tasks. Select one exact taskId and call cognitivedb_ingestion_task_show before a lifecycle mutation.

cognitivedb_ingestion_task_offline

ChatGPT
Take an ONLINE Cloud or Database Ingestion task OFFLINE. This does not delete its saved configuration or cancel any active execution. Pass an exact taskId from cognitivedb_ingestion_task_list or cognitivedb_ingestion_task_show. The Tool reads Task Detail, infers the ingestion type, and accepts only FILE_INGESTION (Cloud) or CONFIGURATION (Database), so callers do not provide an ingestion type. If the task is already OFFLINE, it is returned unchanged without another status request; other states and DAG tasks are rejected. Use cognitivedb_ingestion_execution_cancel separately when an active execution must stop.

cognitivedb_ingestion_task_publish

ChatGPT
Publish an OFFLINE Cloud or Database Ingestion task to ONLINE. Pass the exact taskId returned by cognitivedb_cloud_ingestion_task_create or cognitivedb_database_ingestion_task_create. The Tool reads Task Detail and accepts only FILE_INGESTION (Cloud) or CONFIGURATION (Database), so callers do not provide an ingestion type. If the task is already ONLINE, it is returned unchanged without another publish request; other states and DAG tasks are rejected. Publishing changes task state but does not execute ingestion.

cognitivedb_ingestion_task_show

ChatGPT
Show the saved configuration of one Cloud or Database Ingestion task. Pass an exact taskId from cognitivedb_ingestion_task_list or a task-create result; the Tool reads Task Detail, infers the ingestion type, and rejects DAG tasks. Returns common schedule and runtime settings plus type-specific source, target, write, mapping, field-selection, and audit details through a credential-free whitelist. Review the result before publishing or executing the task. This Tool is read-only.

cognitivedb_knowledge_context_get

ChatGPT
Inspect the selected Compute's knowledge context for the current platform Space and authenticated SQL user. This checks the Space's metastore binding; it does not create or switch Spaces. Returns decoded result data when ready; otherwise use cognitivedb_knowledge_execution_result_get with the returned executionId.

cognitivedb_knowledge_entries_list

ChatGPT
List a page of user directories and published documents. Internal assets are excluded. Omit path for the root; assets and its descendants cannot be listed. Continue with every non-null next_page_token in the same scope, even after an empty page. Returns decoded result data when ready; otherwise use cognitivedb_knowledge_execution_result_get with the returned executionId.

cognitivedb_knowledge_execution_result_get

ChatGPT
Retrieve the decoded result or image of an existing knowledge SQL execution without resubmitting it. For uploads this retrieves the task handle; use cognitivedb_knowledge_upload_status_get to check document publication. For a pending execution, wait at least 3 seconds between checks and increase the interval. Stop automatic polling after 20 checks and resume only when the user asks.

cognitivedb_knowledge_file_read

ChatGPT
Read processed document text by chunk, line, or all. PDF/Office sources return analyzed text. Results are limited to 8 MiB of JSON; use chunk/line reads for larger files. Pass each returned image's path and image_id as path and imageId to cognitivedb_knowledge_image_read. Returns decoded result data when ready; otherwise use cognitivedb_knowledge_execution_result_get with the returned executionId.

cognitivedb_knowledge_file_stage_upload

ChatGPT
External only. Stage a non-empty PDF, JPG/JPEG, PNG, DOCX, PPTX, XLS/XLSX/XLSB, or UTF-8 MD/Markdown file up to 67,108,864 bytes (64 MiB), subject to the Space's lower size and allowed-format limits. Returns source, path and tempFileId for cognitivedb_knowledge_upload and later cleanup.

cognitivedb_knowledge_image_read

ChatGPT
Read a published extracted image using its document path and document-relative imageId (images/...). Returns MCP image content with metadata, up to 32 MiB per image. Requires volume READ. Returns decoded result data when ready; otherwise use cognitivedb_knowledge_execution_result_get with the returned executionId.

cognitivedb_knowledge_upload

ChatGPT
Upload an existing S3 object into an existing knowledge volume. Supply compute, catalog, volume and source; schema defaults to main, path to the source filename, and a stable retry key is generated. Cognitive Database determines size and checksum; contentType defaults from the destination extension. Requires volume MODIFY, parent USAGE, and SQL-session S3 read access. Retains the original S3 object. Requires confirmation; overwrite deletes the old document first and cannot be rolled back. SQL success only returns an upload task: use cognitivedb_knowledge_upload_status_get with its task_id until publication. Repeat the same arguments after an uncertain outcome. For changed bytes at the same source URI, override idempotencyKey with a new key. Returns decoded result data when ready; otherwise use cognitivedb_knowledge_execution_result_get with the returned executionId.

cognitivedb_knowledge_upload_status_get

ChatGPT
Get one private upload task for the current uploader and volume; requires MODIFY and parent USAGE. Does not resubmit the upload. The document is published only when task status is SUCCEEDED or external_result is UPLOAD_SUCCEEDED. Wait at least 3 seconds between checks, increase the interval, and stop automatic polling after 20 checks or when user action is required. Returns decoded result data when ready; otherwise use cognitivedb_knowledge_execution_result_get with the returned executionId.

cognitivedb_knowledge_volume_get

ChatGPT
Get knowledge-volume details and a page of user directories and published documents; internal assets are excluded. Continue with every non-null next_page_token, including after an empty page. Returns decoded result data when ready; otherwise use cognitivedb_knowledge_execution_result_get with the returned executionId.

cognitivedb_knowledge_volume_list

ChatGPT
List visible knowledge volumes in a Catalog and Schema. Volumes are schema-level peers of tables. Returns decoded result data when ready; otherwise use cognitivedb_knowledge_execution_result_get with the returned executionId.

cognitivedb_local_import_execute

ChatGPT
Submit the exact TDS fileRef returned by a successful file prepare operation for import into a Cognitive Database table and return an executionId; then call cognitivedb_local_import_execution_wait. Reuse the complete prepared column list with explicit include flags, provide the import timezone, and, for CSV, reuse the effective csvOptions from the corresponding prepare result. DROP_CREATE deletes the existing table and its data.

cognitivedb_local_import_execution_cancel

ChatGPT
Request cancellation of one active Local Import execution. Pass an exact executionId returned by cognitivedb_local_import_execute or cognitivedb_local_import_execution_list. The Tool first reads the latest Run Detail and submits cancellation only for PENDING, DISPATCHING, or RUNNING; an already CANCELLED execution is returned unchanged, and every other status is rejected. A successful mutation returns CANCEL_REQUESTED, which means cancellation was requested, not that the execution has reached CANCELLED; call cognitivedb_local_import_execution_wait to verify the final state. Cancellation does not delete the source file, target table, execution record, or logs, and partial target writes may remain. Do not retry automatically when the outcome is unknown; inspect the same execution first because cancellation may already have been accepted.

cognitivedb_local_import_execution_list

ChatGPT
List historical Local Import executions in the current Workspace using Studio server-side pagination. Use optional searchText to search source files and studioStatus to filter by an exact Studio execution status. Returns safe execution summaries with public executionId, normalized status, original Studio status, source file, target, row count, timestamps, and sanitized failure reason. Continue while hasMore is true. Pass an active executionId to cognitivedb_local_import_execution_wait when it must be awaited. This Tool is read-only.

cognitivedb_local_import_execution_logs

ChatGPT
List a bounded snapshot of structured logs for one Local Import execution. Pass an exact executionId returned by cognitivedb_local_import_execute or cognitivedb_local_import_execution_list. Studio returns the complete log list, so level filtering and page/pageSize pagination are applied by MCP while preserving Studio order. Each message is redacted while preserving line structure and limited to 2048 characters with truncation metadata. Logs may continue to grow while the execution is active. This Tool is read-only and does not poll, wait, cancel, rerun, or download the raw log file.

cognitivedb_local_import_execution_rerun

ChatGPT
Restart one historical Local Import execution using the same executionId. Pass an exact executionId returned by cognitivedb_local_import_execute or cognitivedb_local_import_execution_list. The Tool first reads the latest Run Detail and allows rerun only for COMPLETED, FINISHED, FAILED, RECOVERABLE, CANCELLED, or PAUSED; active and UNKNOWN executions are rejected. Studio reuses the historical file and complete import configuration, including target, columns, parsing, audit fields, compute, and table-exists behavior; use file upload and execute instead when any configuration must change. Studio reuses the same public execution ID and record instead of creating a separate public Execution; its status, logs, rows written, and results may change. The result returns executionReused=true and the same executionId. Each call may append duplicate data or delete and recreate an existing table when the historical behavior was DROP_CREATE. It does not upload the file again. This operation is not idempotent. Do not retry automatically when the outcome is unknown; wait for and inspect the same execution first.

cognitivedb_local_import_execution_wait

ChatGPT
Wait for the Local Import execution returned by cognitivedb_local_import_execute and return its normalized terminal status, target, rows written, and failure reason.

cognitivedb_local_import_file_prepare

ChatGPT
External only. Prepare a non-empty .csv or .parquet file up to 52,428,800 bytes (50 MiB). Returns a temporary fileRef, columns, sample rows and effective CSV options for cognitivedb_local_import_execute.

cognitivedb_mysql_driver_list

ChatGPT
List available MySQL JDBC driver versions and their supported server versions from Studio. Read-only; no database credentials required.

cognitivedb_notebook_runtime_create

ChatGPT
Create the Compute shown as Notebook Runtime in the UI. Clarify ambiguous 'create Compute' requests. CPU C0 / GPU G2; ask for SQL mode. CPU-only engine options. Does not start it. If binding fails, use cognitivedb_notebook_runtime_sql_bind without recreating. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_notebook_runtime_sql_bind

ChatGPT
Associate an existing CPU/GPU Notebook Runtime with a SQL Warehouse, including recovery after creation succeeded but binding failed. Confirms the binding; creates or starts no Compute. The Notebook SQL selection follows this association. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_ontology_create

ChatGPT
Build an initial ontology draft from data sources or extend a non-partitioned ontology incrementally. Use published baselines until the build API supports draft baselines. Find Connector names with cognitivedb_database_connection_list or cognitivedb_cloud_storage_connection_list. Returns the asynchronous build job. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_ontology_delete

ChatGPT
Execute a reviewed deletion plan after confirmation. Returns a job and leaves source data intact. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_ontology_deletion_plan_create

ChatGPT
Preview removal of tables or namespaces from an ontology. Returns an immutable plan and conflicts for review. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_ontology_deletion_plan_get

ChatGPT
Inspect a saved deletion plan and its current status and jobId. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_ontology_draft_get

ChatGPT
Read an immutable draft before publication. Returns definitions, validation, coverage and publication status. Distinguish business objects from document and structural projections. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_ontology_draft_list

ChatGPT
List the drafts of one ontology. Draft revisions are distinct from published versions. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_ontology_job_get

ChatGPT
Read an Ontology job's status, progress and result. For queued/running jobs, poll the same jobId at least 30 seconds apart for builds and 15 seconds apart for updates. Stop on a terminal state or after 20 checks. Report validation and coverage. Do not automatically retry failed or unknown mutations. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_ontology_list

ChatGPT
List accessible ontologies and published versions. Request overview to inspect saved definitions. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_ontology_publish

ChatGPT
Publish a reviewed draft after confirmation. Returns a release job to verify publication. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_ontology_update

ChatGPT
Apply confirmed text to an ontology. Returns a job that prepares a draft or reports NO_CHANGE. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_organization_billing_overview_get

ChatGPT
Get balance, trial grants, current UTC month spend summary and daily FREE-usage limit. For the selected Space's organization; MCP resolves the organization from Space metadata. Billing checks BILLING_CENTER on every read. Exact decimal monetary strings. Authorization or upstream failure fails the entire call; no partial success or database snapshot guarantee. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_organization_daily_usage_get

ChatGPT
Get daily organization costs and usage. Usage is raw quantity, not money or a common unit across resource types. Defaults: today UTC and 30 days earlier; inclusive dates at most 90 days apart. Billing verifies the authenticated user's organization and BILLING_CENTER permission. Queries the selected Space's organization. Monetary amounts are exact decimal strings. Absent values are null. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_organization_monthly_spend_get

ChatGPT
Get monthly organization spend, zero-filled oldest first. Defaults to six months, maximum 36. Billing verifies the authenticated user's organization and BILLING_CENTER permission. Queries the selected Space's organization. Monetary amounts are exact decimal strings. Absent values are null. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_organization_payment_history_list

ChatGPT
List organization funding/payment history. Defaults: page 1, size 20 (maximum 100). Out-of-range pages clamp to the last available page. Billing verifies the authenticated user's organization and BILLING_CENTER permission. Queries the selected Space's organization. Monetary amounts are exact decimal strings. Absent values are null. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_organization_payment_orders_list

ChatGPT
List existing organization payment orders with status filtering and opaque cursor pagination; excludes checkout handles. Billing verifies the authenticated user's organization and BILLING_CENTER permission. Queries the selected Space's organization. Monetary amounts are exact decimal strings. Absent optional fields are omitted, matching the customer payment views. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_organization_payment_refunds_list

ChatGPT
List the organization's latest 100 existing refunds. Does not create refunds; no additional pagination/export capability. Billing verifies the authenticated user's organization and BILLING_CENTER permission. Queries the selected Space's organization. Monetary amounts are exact decimal strings. Absent optional fields are omitted, matching the customer payment views. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_organization_workspace_spend_get

ChatGPT
Get paginated organization costs by Space, including service breakdown. Same UTC date defaults and 90-day difference limit; does not grant access to Space business data. Billing verifies the authenticated user's organization and BILLING_CENTER permission. Queries the selected Space's organization. Monetary amounts are exact decimal strings. Absent values are null. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_owner_transfer

ChatGPT
Transfer a Catalog, Schema, or Table to a new owner using an explicit resource_type and exact resource-name hierarchy. When transferring multiple levels to the same owner, transfer them in Catalog, Schema, then Table order so the target owner receives parent access before child ownership. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_query_history_list

ChatGPT
List a cursor-paginated slice of query history in the current space. Defaults to 20 records when limit is omitted; set limit to request a different page size. Does not automatically fetch subsequent pages. compute, status, and statementType accept a string or a non-empty array. Values within one filter use OR; different filters use AND at the batch level. Status and statement type values are normalized to uppercase. Optionally filter by executedBy, an array of exact executedBy values from history records (not executedByUserId). Requires QUERY_MONITORING permission; when supplied, checks capabilities on each call and reports an error if unavailable or denied. Omit executedBy to use the existing visibility scope. Keep the same filters when following a cursor. Records appear only in page.content, with page.hasMore and page.nextCursor for pagination. SQL previews are limited to 240 characters and marked with sqlPreviewTruncated=true when shortened. Use cognitivedb_query_history_show for statement summaries, then specify statementIndex for single-statement detail. Payload usage and quota bytes are omitted; batch limit status and reasons are retained.

cognitivedb_query_history_show

ChatGPT
Show query history with locally paginated statement summaries (page=1, size=20, maximum size=100). SQL previews contain at most 240 characters and are marked sqlPreviewTruncated=true when shortened. Top-level pagination counts statement records, not result rows. Set statementIndex to an actual returned index for one statement's full diagnostics; do not combine it with page or size. SQL up to 32 KiB UTF-8 is returned as sqlText; larger SQL returns sqlPreview, sqlBytes, and sqlPreviewTruncated=true instead, without SQL text pagination. Previews may contain incomplete SQL: do not execute or save them as full SQL. Current source files may differ from historical SQL. Batch usage, quotas, timings, and counts are retained; batch sqlPreview is omitted. Each call fetches the full batch from Console without caching.

cognitivedb_s3_connection_generate_trust_policy

ChatGPT
Prepare or repair AWS IAM Role trust settings for an S3 connection. Use this tool when no External ID was supplied, or when an AssumeRole, Principal, trust-policy, or External ID configuration must be repaired. If a valid External ID is already available and no trust-related failure exists, call cognitivedb_s3_connection_pretest directly. Reuse a supplied externalId exactly, or omit it to generate one securely. After the user applies the returned AWS configuration, call cognitivedb_s3_connection_pretest; do not create the connection until pretest succeeds.

cognitivedb_schema_permissions_get

ChatGPT
Get the current user's effective capabilities for a Schema, including preview, edit, delete, ownership transfer, and permission management. Only returns capabilities for the current user.

cognitivedb_schema_permissions_grant

ChatGPT
Grant selected Schema privileges to a grantee. Check with the user that the Schema, grantee, and selected privileges are correct before executing. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_schema_permissions_revoke

ChatGPT
Revoke selected direct Schema privileges from a grantee. Check with the user that the Schema, grantee, and selected privileges are correct before executing. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_space_ai_usage_get

ChatGPT
Get a page of AI credit usage for the current space and inclusive UTC date range. Groups use source and model; newapi_token and xclaw_session use source type, owner and model, with the owner UUID as srcId and a configured srcName (default Daro). Ownerless NewAPI/Xclaw rows keep their source type and model with an empty srcId; only an empty srcId together with null srcType marks unattributed usage. Search matches the displayed name or srcId. The difference between from and to must not exceed 90 days (up to 91 calendar dates). Filters apply to AI rows; filterOptions and hasUnknownOwner describe the unfiltered window. This live report can differ temporarily from rolled-up daily credits. Credits and raw tokens are metered units, not money. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_space_current_get

ChatGPT
Get details of the current space selected by X-Workspace-Id. Returns project_id, name, metastore_id, cloud, region, creator, description, member_count, console_hook_done, created_at_unix, and updated_at_unix.

cognitivedb_space_custom_role_create

ChatGPT
Create a custom role in the current space. Use only valid permission codes. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_space_custom_role_delete

ChatGPT
Delete a custom role from the current space. Use the exact role name returned by cognitivedb_space_roles_list. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_space_list

ChatGPT
List spaces visible to the current user.

cognitivedb_space_member_invite_cancel

ChatGPT
Cancel the latest pending invitation for an exact email address in the current space. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_space_member_invite_resend

ChatGPT
Resend the latest pending invitation for an exact email address in the current space. Confirm the email address with the user before calling this tool. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_space_member_remove

ChatGPT
Remove a member from the current space. Use username returned by cognitivedb_space_members_list. Check with the user that the exact member email is correct before executing. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_space_member_role_set

ChatGPT
Set a current space member's role. Use username returned by cognitivedb_space_members_list. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_space_members_invite

ChatGPT
Invite one or more users to the current space by email. A successful call sends invitation emails. The role defaults to Member; Owner cannot be assigned. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_space_members_list

ChatGPT
List members in the current Space selected by X-Workspace-Id. Optionally filter by a case-insensitive email, username, or full name substring and by an exact Space role name. Full results return usernames that can be used with cognitivedb_space_member_role_set.

cognitivedb_space_roles_list

ChatGPT
List built-in and custom roles in the current space, including their permissions, a role field accepted by role mutation tools, and the permissionCodes currently available for custom roles.

cognitivedb_space_usage_get

ChatGPT
Get daily Platform/AI credits for the whole current space and a page of compute metering for the inclusive UTC date range. The difference between from and to must not exceed 90 days (up to 91 calendar dates). Optional filters apply only to computeMetering; dailyCredits stays a whole-space total. Credits are metered units, not money. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_sql_ddl_execute

ChatGPT
Execute one supported DDL statement through Console's durable asynchronous SQL API after user approval.

cognitivedb_sql_dml_execute

ChatGPT
Execute one INSERT, UPDATE, DELETE, or MERGE statement through Console's durable asynchronous SQL API after user approval.

cognitivedb_sql_example_list

ChatGPT
List read-only example SQL files that can be opened or saved as personal files. Defaults to page=1, size=20 (maximum 100). Filter by case-insensitive file-name substring keyword and exact folder; both filters must match when supplied. Blank filters are ignored. Filtering precedes local pagination and preserves Console order. Each call fetches the full example list again; no cache. pagination.total is the filtered count; use pagination.nextPage with the same size and filters to continue. Keep name/folder for show or save-as. Source is reported once at the top level, not repeated on each file.

cognitivedb_sql_example_save_as

ChatGPT
Save an example SQL file as a new personal SQL file, preserving its variables. Returns file.id/name, optional file.type and a short operation result, without content, variables, or environment details. Use cognitivedb_sql_file_show to read the saved file. An invalid write response means the outcome is unconfirmed; inspect the file before retrying.

cognitivedb_sql_example_show

ChatGPT
Show an example SQL file's content and variable definitions. Omit id/computeId/catalog/schema only when null or empty strings; non-empty values and other detail fields are preserved. Content up to 32 KiB UTF-8 is returned in full. Larger files return contentPreview instead of content, with contentBytes and contentTruncated=true; no pagination. The preview may contain incomplete SQL: do not execute it or use it as replacement file content. Open the full file in the editor's SQL Files panel.

cognitivedb_sql_execution_cancel

ChatGPT
Idempotently request cancellation of a durable SQL execution. Acceptance does not guarantee terminal cancellation; use cognitivedb_sql_execution_status_get until CANCELED.

cognitivedb_sql_execution_results_get

ChatGPT
Get a SQL execution result. On the first and normal call, pass only executionId and omit responseMode so the server automatically selects full or summary output. If summary output is selected, follow the response message. Set responseMode to full only when the user explicitly requires complete rows for the selected scope. Never use responseMode to request a subset; use statementIndexes to select one or more statements, or omit it or pass an empty array to include all statements. Incomplete statements may be returned with status and no rows. Failed statements include errorDetail.message when Console provides SQL diagnostics.

cognitivedb_sql_execution_status_get

ChatGPT
Get SQL execution status and progress. For the current running statement, when cacheDiagnostics.isLoading is true, data synchronization is in progress and the SQL may execute more slowly during synchronization. Then wait at least 3 seconds between status checks. Stop automatic polling after 20 status checks, and resume checks only when the user asks.

cognitivedb_sql_file_batch_delete

ChatGPT
Delete multiple personal SQL files in one operation. All targets are resolved before deletion, then deduplicated by file ID in first-seen order. Returns deleted[].id/name, count and result. Count is the number of unique resolved targets, not a backend-reported affected-row count.

cognitivedb_sql_file_copy

ChatGPT
Copy a personal SQL file, preserving its content, variable definitions, and saved computeId, catalog and schema. Returns file.id/name, optional file.type and a short operation result, without content, variables, or environment details. Use cognitivedb_sql_file_show to read the saved file. An invalid write response means the outcome is unconfirmed; inspect the file before retrying.

cognitivedb_sql_file_create

ChatGPT
Create a personal SQL file with optional parameter definitions. Returns file.id/name, optional file.type and a short operation result, without content, variables, or environment details. Use cognitivedb_sql_file_show to read the saved file. An invalid write response means the outcome is unconfirmed; inspect the file before retrying.

cognitivedb_sql_file_delete

ChatGPT
Delete one personal SQL file. Returns deleted[].id/name, count and result. Count is the number of unique resolved targets, not a backend-reported affected-row count.

cognitivedb_sql_file_execute

ChatGPT
Resolve variables, split and validate every statement in a personal SQL File, confirm the exact snapshot, and submit all statements as one durable execution.

cognitivedb_sql_file_favorite

ChatGPT
Favorite a personal SQL file.

cognitivedb_sql_file_list

ChatGPT
List one page of the current user's personal SQL files, preserving folder, favorite state and timestamps. Defaults to page 1 and size 20; maximum size is 100. keyword matches file names case-insensitively. Backend order is preserved. Use pagination.nextPage with the same size and keyword to continue when hasMore is true. Each call fetches the full current list and filters/pages it locally without caching; concurrent file changes can shift pages. Source is reported once at the top level for single-source lists; legacy non-personal source calls remain unpaginated and reject page/size/keyword.

cognitivedb_sql_file_rename

ChatGPT
Rename a personal SQL file. Returns file.id/name, optional file.type and a short operation result, without content, variables, or environment details. Use cognitivedb_sql_file_show to read the saved file. An invalid write response means the outcome is unconfirmed; inspect the file before retrying.

cognitivedb_sql_file_show

ChatGPT
Show a personal SQL file's content and variable definitions. Content up to 32 KiB UTF-8 is returned in full. Larger files return contentPreview instead of content, with contentBytes and contentTruncated=true; no pagination. The preview may contain incomplete SQL: do not execute it or use it as replacement file content. Open the full file in the editor's SQL Files panel.

cognitivedb_sql_file_unfavorite

ChatGPT
Unfavorite a personal SQL file.

cognitivedb_sql_file_update

ChatGPT
Update a personal SQL file's content or variable definitions. Use rename for its name. Preserves its saved computeId, catalog and schema. Returns file.id/name, optional file.type and a short operation result, without content, variables, or environment details. Use cognitivedb_sql_file_show to read the saved file. An invalid write response means the outcome is unconfirmed; inspect the file before retrying.

cognitivedb_sql_file_upload

ChatGPT
External only. Create a personal SQL file from an uploaded UTF-8 .sql file. Maximum content length: 1,048,576 character units (BMP characters count as 1; supplementary characters count as 2). File size must also be ≤4 MiB (4,194,304 bytes). Split oversized scripts only at safe SQL statement boundaries. Supports optional parameter definitions. Returns file.id/name and a short operation result.

cognitivedb_sql_query_execute

ChatGPT
Execute one read-only SQL query through Console's durable asynchronous SQL API. Returns the result when it finishes quickly; otherwise returns executionId for the SQL execution management Tools.

cognitivedb_table_permissions_get

ChatGPT
Get the current user's effective capabilities for a Table, including preview, edit, delete, ownership transfer, and permission management. Only returns capabilities for the current user.

cognitivedb_table_permissions_grant

ChatGPT
Grant selected Table privileges to a grantee. Check with the user that the Table, grantee, and selected privileges are correct before executing. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_table_permissions_list

ChatGPT
List direct and inherited permissions for a Table.

cognitivedb_table_permissions_revoke

ChatGPT
Revoke selected direct Table privileges from a grantee. Check with the user that the Table, grantee, and selected privileges are correct before executing. Requires permission for the requested operation and target resource. Permissions are checked at execution time; listing this tool does not guarantee access.

cognitivedb_user_current_get

ChatGPT
Get the current authenticated user's own profile, including identity, contact details, language and timezone. Does not require a Space or organization and does not return roles or Metastore capabilities.

Capabilities

Writes

Example Prompts

Click any prompt to copy it.

Similar Apps

Listed on MCP App Store

Listed on MCP App Store

Add this badge to your site or README. It links to this page.

HTML

<a href="https://mcp-app-store.keenethics-labs.com/app/cognitive-database"><img src="https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg" alt="Listed on MCP App Store" width="220" height="44"></a>

Markdown

[![Listed on MCP App Store](https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg)](https://mcp-app-store.keenethics-labs.com/app/cognitive-database)