mcp-server-db
This server is a universal MCP gateway that lets AI assistants query and manage multiple database engines (Postgres, Redis, Elasticsearch, MySQL, MongoDB, LDAP) through a unified set of 113+ tools.
Multi-database connectivity: Connect simultaneously to PostgreSQL, Redis, Elasticsearch, MySQL, MongoDB, and LDAP using environment variables or a YAML config.
Read-only querying: Run SELECT/SHOW/EXPLAIN queries and inspect schemas, tables, indexes, views, constraints, and statistics.
Write operations when allowed: Execute inserts, updates, deletes, DDL, Redis mutations, Elasticsearch indexing, MongoDB document operations, and LDAP directory changes (blocked per-connection in read-only mode).
Database introspection and diagnostics: List tables, schemas, indexes, triggers, routines, functions, sequences, views, materialized views, and DDL definitions.
Performance analysis: EXPLAIN query plans, find slow/long-running queries, unused/duplicate/missing indexes, lock info, bloat estimates, and hypothetical index simulations.
Operational control: Monitor connection status, reload config hot, kill queries, run VACUUM/ANALYZE, refresh materialized views, and check cluster health.
Data exploration and seeding: Query documents in MongoDB with filters/aggregations, search Elasticsearch with Query DSL, inspect Redis keys/hashes/sets/lists, and bulk-load sample data into Redis, Elasticsearch, MongoDB, or LDAP.
Safety and resilience: Per-connection read-only enforcement, result truncation caps, lazy connections, retries, circuit breakers, and graceful failure handling.
Provides tools for cluster health checks, index management, search with Query DSL, document retrieval, and bulk indexing operations.
Provides tools for executing SQL queries, extracting DDL table definitions, managing indexes and triggers, with session-level read-only enforcement.
Provides tools for aggregation pipelines, distinct queries, index management, and execution plan explain operations.
Provides tools for querying, DDL extraction, index management, processlist inspection, and session-level read-only enforcement.
Provides tools for querying, schema and index inspection, performance monitoring, database health diagnostics, and session-level read-only enforcement.
Provides tools for key-value operations, hash, set, list management, TTL control, and read-only mode enforcement.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@mcp-server-db list all tables in my PostgreSQL database"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
mcp-server-db
mcp-server-db is a universal, production-grade Model Context Protocol (MCP) server that empowers AI assistants to interact with Postgres, Redis, Elasticsearch, MySQL, MongoDB, and LDAP through a unified suite of 113+ high-performance tools.
Built with resilience in mind, it features background lazy connection management, exponential backoff retries, circuit breaker patterns, strict read-only security enforcement, and on-demand dynamic driver loading.
Why mcp-server-db?
Multi-Database Support: Connect to Postgres, Redis, Elasticsearch (v7, v8 & v9), MySQL, MongoDB, and LDAP directories simultaneously.
On-Demand Light Drivers: Zero bloat. Drivers (
pg,ioredis,mongodb,mysql2,@elastic/elasticsearch,ldapts) are dynamically loaded only when a connection to that database engine is initialized.Ironclad Safety: Independent per-connection read-only mode with session-level enforcement for relational databases and query result truncation guards to prevent LLM context overflows.
Fault-Tolerant Architecture: Server starts instantly regardless of database availability. Dead or unreachable databases fail gracefully without crashing the MCP process or blocking other connections.
Multi-Connection Routing: Query multiple environments (e.g., production read-only replica + local scratch cache) in parallel within the same session.
Related MCP server: Legion Database MCP Server
Supported Database Engines
Database Engine | Default Driver | Supported Versions | Key Features & Guardrails |
PostgreSQL |
| 10+ | Table/Index stats, Health diagnostics, Session-level |
Redis |
| 5+ | Key/String/Hash/Set/List operations, Automatic 1000-item result truncation cap |
Elasticsearch |
| 7.x, 8.x, 9.x | Dual-API version facade (v7 legacy & v8/v9 modern), Cluster health, Query DSL |
MySQL / MariaDB |
| 5.7+, 8.0+ | DDL extraction ( |
MongoDB |
| 4.4+ | Aggregation pipelines, Distinct queries, Index management, Execution plan explain |
LDAP |
| v3 directories (OpenLDAP, AD, etc.) | Search/compare, Add/modify/delete entries, Bulk entry seeding |
Quick Start
1. Installation & Optional Drivers
When installed globally (npm install -g mcp-server-db) or run via npx -y mcp-server-db, npm automatically includes all supported database drivers (pg, ioredis, @elastic/elasticsearch, mysql2, mongodb, ldapts) out of the box via optionalDependencies.
# Global installation (includes all drivers out of the box)
npm install -g mcp-server-db
# Local project installation
npm install mcp-server-dbIf you install mcp-server-db locally using --no-optional to keep your node_modules lightweight, you can install only the specific drivers you need:
# PostgreSQL
npm install pg
# Redis
npm install ioredis
# Elasticsearch
npm install @elastic/elasticsearch # For v8 & v9
npm install es7-client # For v7 legacy
# MySQL / MariaDB
npm install mysql2
# MongoDB
npm install mongodb
# LDAP
npm install ldapts2. Zero-Install Execution via NPX
Add mcp-server-db directly to your MCP client config file (e.g. Claude Desktop, Claude Code, Gemini CLI, Cursor):
{
"mcpServers": {
"mcp-server-db": {
"type": "stdio",
"command": "npx",
"args": ["-y", "mcp-server-db"],
"env": {
"POSTGRES_URL": "postgresql://user:password@localhost:5432/mydb",
"REDIS_URL": "redis://localhost:6379",
"MONGODB_URL": "mongodb://localhost:27017",
"LDAP_URL": "ldap://localhost:389",
"LDAP_BIND_DN": "cn=admin,dc=example,dc=com",
"LDAP_BIND_PASSWORD": "admin-password"
}
}
}
}3. Fast Setup with Environment Variables
Define connection strings in env to start immediately without a configuration file:
POSTGRES_URL=postgresql://user:password@localhost:5432/mydb
REDIS_URL=redis://localhost:6379
ELASTICSEARCH_URL=http://localhost:9200
MYSQL_URL=mysql://user:password@localhost:3306/mydb
MONGODB_URL=mongodb://localhost:27017
LDAP_URL=ldap://localhost:389
LDAP_BIND_DN=cn=admin,dc=example,dc=com
LDAP_BIND_PASSWORD=admin-password
POSTGRES_READ_ONLY=true
REDIS_READ_ONLY=true
MYSQL_READ_ONLY=true
MONGODB_READ_ONLY=true
LDAP_READ_ONLY=trueComplete Tools Reference (113 Tools)
System & Management Tools (2)
db_status: Inspect connection states (idle, connecting, connected, retrying, failed, circuit_open), read-only mode, and error logs across all configured databases. Supports filtering byconnectionId(orid) to check a single database, filtering bytype(postgres/redis/etc.), and active live probing (probe: true) with round-trip latency measurement.db_reload_config: Hot-reloaddatabases.config.ymlfrom disk and reconcile connections dynamically without interrupting active queries.
Postgres Tools (30)
pg_query: Execute read-only SELECT queries (enforces single-statement extended query protocol).pg_execute_sql: Execute write SQL statements (INSERT, UPDATE, DELETE, DDL). Blocked in read-only mode.pg_list_tables: List tables in a target schema.pg_describe_table: Inspect column data types, ordinal position, and nullability.pg_list_schemas: List all database schemas.pg_list_views: List all views in a schema.pg_list_indexes: List indexes and index definitions for a table.pg_list_triggers: List triggers and action statements in a schema.pg_table_stats: Get total size, table size, index size, and estimated row count.pg_list_constraints: List primary keys, foreign keys, and check constraints.pg_explain_query: Inspect query execution plans with EXPLAIN.pg_active_queries: Retrieve running non-idle queries and durations frompg_stat_activity.pg_list_functions: List user-defined functions, arguments, and return types.pg_list_sequences: List sequence objects in a schema.pg_database_info: Get database name, current user, PostgreSQL version, and database size.pg_kill_query: Terminate a running backend process by PID. Blocked in read-only mode.pg_vacuum_analyze: Run VACUUM ANALYZE to update statistics and reclaim dead tuples. Blocked in read-only mode.pg_list_materialized_views: List materialized views in a schema.pg_refresh_materialized_view: Refresh a materialized view (supports CONCURRENTLY). Blocked in read-only mode.pg_index_usage: Inspect index scan counts and tuple fetch statistics.pg_lock_info: Retrieve active database locks and query durations.pg_get_top_queries: Get slow queries frompg_stat_statements.pg_explain_hypothetical_index: Simulate hypothetical index performance usinghypopg.pg_database_health: Run health audit (connections utilization, buffer cache hit ratio, dead rows, replication lag).pg_sizes_overview: Breakdown storage sizes for database, tables, and indexes.pg_unused_indexes: Identify zero-scan candidate indexes to drop.pg_duplicate_indexes: Find overlapping or redundant indexes.pg_long_running_queries: Find queries exceeding a duration threshold.pg_missing_indexes: Identify high-sequential-scan tables needing indexes.pg_bloat_estimate: Estimate table bloat ratio based on dead tuple count.
Redis Tools (27)
redis_get: Get string value at key.redis_set: Set key value. Blocked in read-only mode.redis_mset: Set multiple key/value pairs in a single command (e.g. for seeding). Blocked in read-only mode.redis_del: Delete key. Blocked in read-only mode.redis_keys: List keys matching a glob pattern (capped at 1000 items).redis_ttl: Get remaining TTL in seconds.redis_hget: Get hash field value.redis_hset: Set hash field value. Blocked in read-only mode.redis_hdel: Delete hash fields. Blocked in read-only mode.redis_hgetall: Get all hash fields and values.redis_hexists: Check hash field existence.redis_sadd: Add member to set. Blocked in read-only mode.redis_srem: Remove member from set. Blocked in read-only mode.redis_smembers: List set members (capped at 1000 items).redis_sismember: Test set membership.redis_lpush: Prepend element to list. Blocked in read-only mode.redis_rpush: Append element to list. Blocked in read-only mode.redis_lpop: Remove and return first list element. Blocked in read-only mode.redis_rpop: Remove and return last list element. Blocked in read-only mode.redis_lrange: Get range of list elements (capped at 1000 items).redis_llen: Get list length.redis_exists: Check key existence.redis_expire: Set key expiration in seconds. Blocked in read-only mode.redis_type: Retrieve data type of key.redis_incr: Increment integer key. Blocked in read-only mode.redis_decr: Decrement integer key. Blocked in read-only mode.redis_flushdb: Clear current Redis database. Blocked in read-only mode.
Elasticsearch Tools (11)
es_cluster_health: Get status (green/yellow/red), node counts, and shard metrics.es_list_indices: List indices, doc counts, store size, and status.es_index_stats: Get detailed metrics for a specific index.es_search: Execute search using Query DSL.es_count: Count documents matching a query.es_get_doc: Fetch document by ID.es_index_doc: Index or overwrite document. Blocked in read-only mode.es_bulk_index: Index multiple documents into one index in a single_bulkrequest, with per-document success/failure reporting (e.g. for seeding). Blocked in read-only mode.es_update_doc: Partially update document fields. Blocked in read-only mode.es_delete_doc: Delete document by ID. Blocked in read-only mode.es_delete_by_query: Delete documents matching Query DSL. Blocked in read-only mode.
MySQL Tools (18)
mysql_query: Execute read-only SELECT/SHOW/EXPLAIN queries.mysql_execute_sql: Execute write SQL (INSERT, UPDATE, DELETE, DDL). Blocked in read-only mode.mysql_list_databases: List databases on server.mysql_list_tables: List tables in database.mysql_list_views: List views and security options.mysql_describe_table: Show column definitions and metadata.mysql_show_create_table: Extract rawCREATE TABLEDDL statement.mysql_show_create_view: Extract rawCREATE VIEWDDL statement.mysql_list_indexes: Show table indexes (SHOW INDEX).mysql_list_triggers: List database triggers.mysql_list_routines: List stored procedures and functions.mysql_list_constraints: List foreign keys and constraints.mysql_table_stats: Retrieve estimated rows, data length, and index size.mysql_active_queries: Retrieve running threads (SHOW FULL PROCESSLIST).mysql_kill_query: Kill running process by ID. Blocked in read-only mode.mysql_explain_query: Analyze query execution plan (EXPLAIN).mysql_global_status: Query server global status metrics.mysql_analyze_table: RunANALYZE TABLEto refresh statistics. Blocked in read-only mode.
MongoDB Tools (19)
mongo_list_databases: List databases on MongoDB cluster.mongo_list_collections: List collections in database.mongo_find: Query documents with filter, projection, sort, limit, and skip.mongo_distinct: Get unique field values.mongo_aggregate: Execute MongoDB aggregation pipeline.mongo_count_documents: Count documents matching filter.mongo_insert_one: Insert single document. Blocked in read-only mode.mongo_insert_many: Insert array of documents. Blocked in read-only mode.mongo_update_one: Update single document matching filter. Blocked in read-only mode.mongo_update_many: Update multiple documents. Blocked in read-only mode.mongo_delete_one: Delete single document. Blocked in read-only mode.mongo_delete_many: Delete multiple documents. Blocked in read-only mode.mongo_create_index: Create collection index. Blocked in read-only mode.mongo_drop_index: Drop collection index by name. Blocked in read-only mode.mongo_list_indexes: List all collection indexes.mongo_db_stats: Get database storage statistics.mongo_collection_stats: Get storage, index size, and object count for collection.mongo_explain: Get query planner execution plan and index scan metrics.mongo_server_status: Get cluster memory, connection, and operation stats.
LDAP Tools (6)
ldap_search: Search a directory by base DN, filter, and scope (capped at 500 entries).ldap_compare: Compare an attribute/value pair against an entry.ldap_add: Create a new entry. Blocked in read-only mode.ldap_add_bulk: Create multiple entries in one call (e.g. for seeding), with per-entry success/failure reporting. Blocked in read-only mode.ldap_modify: Add, replace, or delete an attribute's values on an entry. Blocked in read-only mode.ldap_delete: Delete an entry. Blocked in read-only mode.
Multi-Connection YAML Configuration
For complex setups with multiple named databases, create config/databases.config.yml (copy config/databases.config.example.yml):
connections:
- id: primary-pg
type: postgres
connectionString: ${POSTGRES_URL}
defaultSchema: public
statementTimeoutMs: 30000
readOnly: true
- id: app-cache
type: redis
connectionString: ${REDIS_URL}
readOnly: false
- id: search-logs
type: elasticsearch
connectionString: ${ELASTICSEARCH_URL}
apiVersion: "9"
readOnly: true
- id: legacy-cluster
type: elasticsearch
connectionString: ${LEGACY_ELASTICSEARCH_URL}
apiVersion: "7"
readOnly: true
- id: analytics-mysql
type: mysql
connectionString: ${MYSQL_URL}
statementTimeoutMs: 30000
readOnly: true
- id: document-store
type: mongodb
connectionString: ${MONGODB_URL}
defaultDatabase: production
readOnly: true
- id: directory
type: ldap
connectionString: ${LDAP_URL}
bindDn: ${LDAP_BIND_DN}
bindPassword: ${LDAP_BIND_PASSWORD}
readOnly: falseLocal Development & Custom Builds
1. Build and Run Locally
git clone https://github.com/Nam088/mcp-database-server-v2.git
cd mcp-database-server
npm install
npm run build # Compiles TypeScript to dist/
npm test # Runs Vitest test suite2. Configure Local Build in MCP Clients
{
"mcpServers": {
"mcp-server-db": {
"type": "stdio",
"command": "node",
"args": ["/absolute/path/to/mcp-database-server/dist/index.js"],
"env": {
"DATABASES_CONFIG_PATH": "/absolute/path/to/mcp-database-server/config/databases.config.yml"
}
}
}
}3. Claude Code CLI
claude mcp add --scope user mcp-server-db node /absolute/path/to/mcp-database-server/dist/index.js --env DATABASES_CONFIG_PATH=/absolute/path/to/mcp-database-server/config/databases.config.ymlClient Config Paths
Claude Desktop:
~/Library/Application Support/Claude/claude_desktop_config.jsonClaude Code (Global):
~/.claude.jsonGemini CLI / IDE:
~/.gemini/config/mcp_config.jsonCursor / Project Root:
.mcp.json
License
Distributed under the MIT License. See LICENSE for details.
Available Tools
113 toolsdb_reload_configA
Reload the databases.config.yml configuration file from disk, recreate connections, and reconnect.
Inputs: no named parameters.
db_statusA
List the live connection state of every configured database (postgres/redis/elasticsearch), including state, readOnly mode, last error, and next retry time.
Inputs: no named parameters.
es_bulk_indexA
Index (create or overwrite) multiple documents into one index in a single request via the Elasticsearch _bulk API (e.g. for seeding test/sample data) — far more efficient than one es_index_doc call per document. Reports per-document success/failure rather than aborting the whole batch on one bad document. Blocked when that connection's own readOnly mode is enabled.
Inputs: index (required), documents (required), connectionId.
es_cluster_healthA
Get Elasticsearch cluster health: status (green/yellow/red), node count, and shard counts.
Inputs: connectionId.
es_countA
Count documents in an index matching a query (or all documents if no query is given).
Inputs: index (required), query, connectionId.
es_delete_by_queryB
Delete all documents in an index matching a query. Blocked when that connection's own readOnly mode is enabled.
Inputs: index (required), query (required), connectionId.
es_delete_docC
Delete a document by id. Blocked when that connection's own readOnly mode is enabled.
Inputs: id (required), index (required), connectionId.
es_get_docB
Get a document by id from an index.
Inputs: id (required), index (required), connectionId.
es_index_docA
Index (create or overwrite) a document. Blocked when that connection's own readOnly mode is enabled.
Inputs: index (required), document (required), id, connectionId.
es_index_statsA
Get stats (doc count, store size, segments) for a specific index.
Inputs: index (required), connectionId.
es_list_indicesB
List all indices with document count, size, and health status.
Inputs: connectionId.
es_searchB
Run a search query against an index using Elasticsearch Query DSL.
Inputs: index (required), size, query, connectionId.
es_update_docB
Partially update a document by id. Blocked when that connection's own readOnly mode is enabled.
Inputs: id (required), doc (required), index (required), connectionId.
ldap_addB
Create a new LDAP entry. Blocked when that connection's own readOnly mode is enabled.
Inputs: dn (required), attributes (required), connectionId.
ldap_add_bulkA
Create multiple LDAP entries in one call (e.g. for seeding test/sample data). LDAP has no native multi-entry add operation, so this adds each entry sequentially and reports per-entry success/failure rather than aborting on the first error — a bad entry never blocks the rest of the batch. Blocked when that connection's own readOnly mode is enabled.
Inputs: entries (required), connectionId.
ldap_compareB
Compare an attribute/value pair against an LDAP entry. Returns true if they match.
Inputs: dn (required), value (required), attribute (required), connectionId.
ldap_deleteB
Delete an LDAP entry. Blocked when that connection's own readOnly mode is enabled.
Inputs: dn (required), connectionId.
ldap_modifyA
Add, replace, or delete a single attribute's values on an existing LDAP entry. Blocked when that connection's own readOnly mode is enabled.
Inputs: dn (required), attribute (required), operation (required), values, +1 more.
ldap_searchA
Search an LDAP directory. Returns matching entries (capped at 500; result is truncated=true with a total count if there are more).
Inputs: baseDn (required), scope, filter, sizeLimit, +2 more.
mongo_aggregateC
Run an aggregation pipeline on a MongoDB collection.
Inputs: pipeline (required), collection (required), database, connectionId.
mongo_collection_statsB
Get detailed storage, index size, and document count statistics for a MongoDB collection.
Inputs: collection (required), database, connectionId.
mongo_count_documentsB
Count documents matching a filter in a MongoDB collection.
Inputs: collection (required), filter, database, connectionId.
mongo_create_indexA
Create an index on a MongoDB collection. Blocked when readOnly mode is enabled.
Inputs: keys (required), collection (required), name, unique, +2 more.
mongo_db_statsC
Get statistics for a MongoDB database.
Inputs: database, connectionId.
mongo_delete_manyB
Delete multiple documents matching filter in MongoDB. Blocked when readOnly mode is enabled.
Inputs: filter (required), collection (required), database, connectionId.
mongo_delete_oneA
Delete a single document matching filter in MongoDB. Blocked when readOnly mode is enabled.
Inputs: filter (required), collection (required), database, connectionId.
mongo_distinctC
Find distinct values for a specified field in a MongoDB collection.
Inputs: field (required), collection (required), filter, database, +1 more.
mongo_drop_indexA
Drop an index by name on a MongoDB collection. Blocked when readOnly mode is enabled.
Inputs: indexName (required), collection (required), database, connectionId.
mongo_explainA
Get execution plan and index scan statistics for a MongoDB find query.
Inputs: collection (required), filter, database, verbosity, +1 more.
mongo_findC
Query documents from a MongoDB collection.
Inputs: collection (required), skip, sort, limit, +4 more.
mongo_insert_manyB
Insert multiple documents into a MongoDB collection. Blocked when readOnly mode is enabled.
Inputs: documents (required), collection (required), database, connectionId.
mongo_insert_oneA
Insert a single document into a MongoDB collection. Blocked when readOnly mode is enabled.
Inputs: document (required), collection (required), database, connectionId.
mongo_list_collectionsB
List all collections in a MongoDB database.
Inputs: database, connectionId.
mongo_list_databasesB
List all databases in MongoDB.
Inputs: connectionId.
mongo_list_indexesC
List all indexes on a MongoDB collection.
Inputs: collection (required), database, connectionId.
mongo_server_statusA
Get MongoDB server status (connections, memory, uptime, operations).
Inputs: connectionId.
mongo_update_manyC
Update multiple documents matching filter in MongoDB. Blocked when readOnly mode is enabled.
Inputs: filter (required), update (required), collection (required), upsert, +2 more.
mongo_update_oneC
Update a single document matching filter in MongoDB. Blocked when readOnly mode is enabled.
Inputs: filter (required), update (required), collection (required), upsert, +2 more.
mysql_active_queriesB
List currently running queries and processes in MySQL.
Inputs: connectionId.
mysql_analyze_tableB
Analyze key distribution for a MySQL table. Blocked when readOnly mode is enabled.
Inputs: table (required), database, connectionId.
mysql_describe_tableB
Show column definitions and metadata for a MySQL table.
Inputs: table (required), database, connectionId.
mysql_execute_sqlB
Run any SQL statement (INSERT/UPDATE/DELETE/DDL) against a configured MySQL connection. Blocked when readOnly mode is enabled.
Inputs: sql (required), connectionId.
mysql_explain_queryB
Get the execution plan for a MySQL query.
Inputs: sql (required), connectionId.
mysql_global_statusC
Show server status variables in MySQL.
Inputs: filter, connectionId.
mysql_kill_queryA
Kill a running query or connection thread by ID in MySQL. Blocked when readOnly mode is enabled.
Inputs: processId (required), connectionId.
mysql_list_constraintsA
List table constraints and foreign keys in a MySQL database.
Inputs: table, database, connectionId.
mysql_list_databasesA
List all databases on a MySQL connection.
Inputs: connectionId.
mysql_list_indexesB
Show all indexes for a specified MySQL table.
Inputs: table (required), database, connectionId.
mysql_list_routinesC
List stored procedures and functions in a MySQL database.
Inputs: database, connectionId.
mysql_list_tablesB
List tables in a MySQL database.
Inputs: database, connectionId.
mysql_list_triggersC
List triggers in a MySQL database.
Inputs: database, connectionId.
mysql_list_viewsC
List views in a MySQL database.
Inputs: database, connectionId.
mysql_queryA
Run a read-only SELECT/SHOW/EXPLAIN query against a configured MySQL connection.
Inputs: sql (required), connectionId.
mysql_show_create_tableB
Get the exact CREATE TABLE DDL statement for a MySQL table.
Inputs: table (required), database, connectionId.
mysql_show_create_viewC
Get the exact CREATE VIEW DDL statement for a MySQL view.
Inputs: view (required), database, connectionId.
mysql_table_statsC
Get size, row count estimates, and index sizes for tables in a MySQL database.
Inputs: database, connectionId.
pg_active_queriesB
Get active queries currently running on the server.
Inputs: connectionId.
pg_bloat_estimateA
Estimate table and index bloat using pg_stat_user_tables statistics (dead tuples vs live tuples ratio). Returns tables where dead tuple ratio exceeds a threshold.
Inputs: schema, connectionId, min_dead_ratio_pct.
pg_database_healthB
Get a comprehensive database health report (connections, buffer cache, invalid indexes, dead rows, and replication).
Inputs: connectionId.
pg_database_infoC
Get metadata and size information about the current database.
Inputs: connectionId.
pg_describe_tableA
Describe the columns of a table in the specified schema of a configured Postgres connection.
Inputs: table (required), schema, connectionId.
pg_duplicate_indexesA
Find redundant indexes where one index's leading columns are a subset of another, making the smaller one unnecessary.
Inputs: schema, connectionId.
pg_execute_sqlA
Run any SQL statement (INSERT/UPDATE/DELETE/DDL) against a configured Postgres connection. Blocked when that connection's own readOnly mode is enabled.
Inputs: sql (required), connectionId.
pg_explain_hypothetical_indexA
Explain a query's execution plan simulating hypothetical indexes using the hypopg extension (if installed).
Inputs: sql (required), indexes (required), connectionId.
pg_explain_queryA
Run EXPLAIN on a read-only SELECT query to analyze query performance.
Inputs: sql (required), connectionId.
pg_get_top_queriesA
Get the slowest SQL queries based on execution time from pg_stat_statements (if the extension is enabled).
Inputs: limit, connectionId.
pg_index_usageA
Retrieve index usage statistics for all user tables in the specified schema.
Inputs: schema, connectionId.
pg_kill_queryB
Terminate an active query by PID. Blocked when that connection's own readOnly mode is enabled.
Inputs: pid (required), connectionId.
pg_list_constraintsA
List constraints (foreign keys, primary keys, check constraints) on a table in the specified schema.
Inputs: table (required), schema, connectionId.
pg_list_functionsA
List all user-defined functions in the specified schema.
Inputs: schema, connectionId.
pg_list_indexesB
List all indexes for a given table in the specified schema.
Inputs: table (required), schema, connectionId.
pg_list_materialized_viewsA
List all materialized views in the specified schema.
Inputs: schema, connectionId.
pg_list_schemasB
List all database schemas.
Inputs: connectionId.
pg_list_sequencesA
List all sequences in the specified schema.
Inputs: schema, connectionId.
pg_list_tablesB
List all tables in the specified schema of a configured Postgres connection.
Inputs: schema, connectionId.
pg_list_triggersB
List all triggers in the specified schema.
Inputs: schema, connectionId.
pg_list_viewsB
List all views in the specified schema.
Inputs: schema, connectionId.
pg_lock_infoB
Get active lock information from the database catalog.
Inputs: connectionId.
pg_long_running_queriesC
Find queries that have been running longer than a given threshold in seconds.
Inputs: connectionId, min_duration_seconds.
pg_missing_indexesA
Find tables that have high sequential scan counts relative to index scans, suggesting they might benefit from new indexes.
Inputs: schema, connectionId, min_seq_scans.
pg_queryA
Run a read-only SELECT query against a configured Postgres connection.
Inputs: sql (required), connectionId.
pg_refresh_materialized_viewB
Refresh a materialized view. Blocked when that connection's own readOnly mode is enabled.
Inputs: view (required), schema, concurrently, connectionId.
pg_sizes_overviewB
Get size statistics for the database and all tables/indexes in a schema.
Inputs: schema, connectionId.
pg_table_statsA
Get statistics for a given table in the specified schema, including table size, index size, and estimated row count.
Inputs: table (required), schema, connectionId.
pg_unused_indexesA
Find indexes that have never been used (zero scans since last stats reset). Great for identifying candidates to DROP to save space and reduce write overhead.
Inputs: schema, connectionId.
pg_vacuum_analyzeA
Run VACUUM ANALYZE on a table to clean dead tuples and update database statistics. Blocked when that connection's own readOnly mode is enabled.
Inputs: table (required), schema, connectionId.
redis_decrB
Decrement the integer value of a key. Blocked when that connection's own readOnly mode is enabled.
Inputs: key (required), connectionId.
redis_delC
Delete a Redis key. Blocked when that connection's own readOnly mode is enabled.
Inputs: key (required), connectionId.
redis_existsC
Check key existence.
Inputs: key (required), connectionId.
redis_expireB
Set a key's time to live in seconds. Blocked when that connection's own readOnly mode is enabled.
Inputs: key (required), seconds (required), connectionId.
redis_flushdbA
Delete all keys from the current database. Blocked when that connection's own readOnly mode is enabled.
Inputs: connectionId.
redis_getB
Get the value stored at a Redis key.
Inputs: key (required), connectionId.
redis_hdelB
Delete one or more hash fields. Blocked when that connection's own readOnly mode is enabled.
Inputs: key (required), field (required), connectionId.
redis_hexistsC
Check if a hash field exists.
Inputs: key (required), field (required), connectionId.
redis_hgetC
Get the value of a hash field.
Inputs: key (required), field (required), connectionId.
redis_hgetallA
Get all fields and values of a hash.
Inputs: key (required), connectionId.
redis_hsetB
Set the value of a hash field. Blocked when that connection's own readOnly mode is enabled.
Inputs: key (required), field (required), value (required), connectionId.
redis_incrA
Increment the integer value of a key. Blocked when that connection's own readOnly mode is enabled.
Inputs: key (required), connectionId.
redis_keysA
List Redis keys matching a glob pattern (e.g. 'user:*'). Capped at 1000 keys; result is truncated=true with a total count if there are more.
Inputs: pattern (required), connectionId.
redis_llenB
Get the length of a list.
Inputs: key (required), connectionId.
redis_lpopC
Remove and return the first element of a list. Blocked when that connection's own readOnly mode is enabled.
Inputs: key (required), connectionId.
redis_lpushA
Prepend a value to a list. Blocked when that connection's own readOnly mode is enabled.
Inputs: key (required), value (required), connectionId.
redis_lrangeB
Get a range of elements from a list. Capped at 1000 elements; result is truncated=true with a total count if there are more.
Inputs: key (required), stop (required), start (required), connectionId.
redis_msetA
Set multiple string keys to their values in a single command (e.g. for seeding test/sample data) — far more efficient than one redis_set call per key. Blocked when that connection's own readOnly mode is enabled.
Inputs: entries (required), connectionId.
redis_rpopA
Remove and return the last element of a list. Blocked when that connection's own readOnly mode is enabled.
Inputs: key (required), connectionId.
redis_rpushB
Append a value to a list. Blocked when that connection's own readOnly mode is enabled.
Inputs: key (required), value (required), connectionId.
redis_saddA
Add a member to a set. Blocked when that connection's own readOnly mode is enabled.
Inputs: key (required), member (required), connectionId.
redis_setB
Set a Redis key to a value. Blocked when that connection's own readOnly mode is enabled.
Inputs: key (required), value (required), connectionId.
redis_sismemberA
Check if a value is a member of a set.
Inputs: key (required), member (required), connectionId.
redis_smembersA
Get all members of a set. Capped at 1000 members; result is truncated=true with a total count if there are more.
Inputs: key (required), connectionId.
redis_sremB
Remove a member from a set. Blocked when that connection's own readOnly mode is enabled.
Inputs: key (required), member (required), connectionId.
redis_ttlA
Get the remaining time to live (in seconds) of a Redis key, or -1 if it has none.
Inputs: key (required), connectionId.
redis_typeC
Retrieve the internal data type of a key.
Inputs: key (required), connectionId.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 113 tools
Engine prefixes and resource-specific predicates clearly separate tools, e.g., pg_list_tables vs redis_hgetall vs mongo_find. Minor overlap exists among PostgreSQL statistics and index analysis tools (pg_table_stats vs pg_sizes_overview vs pg_database_health), but descriptions generally differentiate them.
The overwhelming majority follow a consistent engine_verb_noun snake_case pattern, such as pg_list_tables, redis_hset, mongo_insert_one, and es_delete_doc. Exceptions like db_status (no verb) and the variation between es_bulk_index and ldap_add_bulk are minor deviations from an otherwise strong convention.
At 113 tools, the surface is far beyond the typical 3-15 tool scope and overwhelms agents even though the server spans multiple database engines. The sheer number exceeds the 'too many' threshold, and the breadth would be better split into dedicated per-engine MCP servers.
CRUD, schema inspection, query execution, and administrative operations are thoroughly covered for PostgreSQL, MySQL, MongoDB, Redis, Elasticsearch, and LDAP. Notable gaps include Redis sorted-set operations, Elasticsearch index creation/deletion, and Mongo drop collection/database, but these are minor relative to the very broad surface.
Maintenance
Related MCP Connectors
PostgreSQL, MySQL, and SQL Server in one session. 26 read-only MCP tools for AI agents.
Text-to-SQL MCP server: read-only queries on PostgreSQL, MySQL and SQL Server with your real schema
MCP server unifying ERPs, CRMs, APIs and knowledge base for Claude, ChatGPT and Gemini.
Hosted SQL Server MCP server for Claude, ChatGPT and Cursor, with each action switched per account.
Related MCP Servers
- AlicenseNot gradedqualityNot gradedmaintenanceA universal database gateway MCP server that enables AI assistants to connect to and query multiple databases (PostgreSQL, MySQL, MariaDB, SQL Server, SQLite) with support for schema exploration, SQL execution, and secure connections via SSH tunnels.17 npm-
- AlicenseNot gradedqualityDmaintenanceA unified MCP server for querying and managing multiple database types (PostgreSQL, MySQL, SQL Server, etc.) via natural language through AI assistants.GPL 3.0
- AlicenseAqualityDmaintenanceA multi-database MCP server that enables LLMs to safely interact with MySQL, PostgreSQL, SQLite, and others through a unified tool interface, with permission modes and schema resources.1015 npm43MIT
- AlicenseAqualityBmaintenanceZero-config MCP server that empowers AI agents to safely query SQL and NoSQL databases like PostgreSQL, MySQL, SQLite, MongoDB, and Redis.547 npm1MIT