PostgreSQL MCP Server
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| OPENAI_MODEL | No | OpenAI model to use | gpt-5.2-mini |
| CACHE_ENABLED | No | Enable schema caching | true |
| DATABASE_HOST | No | PostgreSQL host | localhost |
| DATABASE_NAME | Yes | Database name | |
| DATABASE_PORT | No | PostgreSQL port | 5432 |
| DATABASE_USER | Yes | Database user | |
| CACHE_MAX_SIZE | No | Maximum number of cached schemas | 100 |
| OPENAI_API_KEY | Yes | OpenAI API key | |
| OPENAI_TIMEOUT | No | API timeout in seconds | 30 |
| CACHE_SCHEMA_TTL | No | Schema cache TTL in seconds | 3600 |
| DATABASE_PASSWORD | Yes | Database password | |
| OPENAI_MAX_TOKENS | No | Maximum tokens per request | 32000 |
| SECURITY_MAX_ROWS | No | Maximum rows per query | 10000 |
| OPENAI_TEMPERATURE | No | Model temperature | 0.0 |
| ADDITIONAL_DATABASES | No | JSON array of additional database configurations | |
| DATABASE_MAX_POOL_SIZE | No | Maximum connection pool size | 20 |
| DATABASE_MIN_POOL_SIZE | No | Minimum connection pool size | 5 |
| RESILIENCE_MAX_RETRIES | No | Maximum retry count | 3 |
| RESILIENCE_RETRY_DELAY | No | Initial retry delay in seconds | 1.0 |
| SECURITY_ALLOW_EXPLAIN | No | Allow EXPLAIN statements | false |
| OBSERVABILITY_LOG_LEVEL | No | Log level | INFO |
| RESILIENCE_RETRY_JITTER | No | Retry delay jitter ratio (0-1) | 0.1 |
| SECURITY_BLOCKED_TABLES | No | Comma-separated list of blocked tables | |
| DATABASE_COMMAND_TIMEOUT | No | Query timeout in seconds | 30 |
| OBSERVABILITY_LOG_FORMAT | No | Log format (json/text) | json |
| SECURITY_BLOCKED_COLUMNS | No | Comma-separated list of blocked columns | |
| RESILIENCE_BACKOFF_FACTOR | No | Exponential backoff factor | 2.0 |
| OBSERVABILITY_METRICS_PORT | No | Metrics HTTP port | 9090 |
| SECURITY_BLOCKED_FUNCTIONS | No | Comma-separated list of blocked PostgreSQL functions | |
| SECURITY_MAX_EXECUTION_TIME | No | Query execution timeout in seconds | 30 |
| OBSERVABILITY_METRICS_ENABLED | No | Enable Prometheus metrics | true |
| SECURITY_ALLOW_WRITE_OPERATIONS | No | Allow INSERT/UPDATE/DELETE operations | false |
| RESILIENCE_LLM_CONCURRENCY_LIMIT | No | Maximum concurrent LLM calls | 5 |
| RESILIENCE_CIRCUIT_BREAKER_TIMEOUT | No | Circuit breaker timeout in seconds | 60 |
| RESILIENCE_QUERY_CONCURRENCY_LIMIT | No | Maximum concurrent queries | 10 |
| RESILIENCE_RATE_LIMIT_WAIT_TIMEOUT | No | Rate limit wait timeout in seconds | 10 |
| RESILIENCE_CIRCUIT_BREAKER_THRESHOLD | No | Failure count before circuit breaker trips | 5 |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": true
} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
| experimental | {
"tasks": {
"list": {},
"cancel": {},
"requests": {
"tools": {
"call": {}
},
"prompts": {
"get": {}
},
"resources": {
"read": {}
}
}
}
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| addA | Add two numbers |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 1 tool
Only one tool exists, so there is no possibility of confusion between tools. However, the tool's purpose (adding two numbers) is completely unrelated to the server's stated domain of PostgreSQL, making the set conceptually incoherent.
With a single tool named 'add', there is no pattern to evaluate. The name is also extremely generic and does not reflect the PostgreSQL domain.
A single trivial arithmetic tool is an extreme mismatch for a PostgreSQL MCP server, which would be expected to expose database operations like querying, schema management, or connection handling.
The tool surface is severely incomplete for the stated purpose. A PostgreSQL server with only 'add' provides no database-related functionality and would fail any real agent task involving PostgreSQL.