Postgres MCP

Safely query PostgreSQL databases to inspect schemas, tables, and data.

Source

Official Registry

Updated

2026-06-11

// securityCaution

Requires keys· MCP_KEY_HOST, MCP_KEY_PORT, MCP_KEY_NAME, MCP_KEY_USER, MCP_KEY_PASS, MCP_KEY_SSLMODE
Runs locally
Official / curated
Open source
Community-trusted

Summary

Overall this is an official-registry, open-source PostgreSQL MCP tool with recent updates, so supply-chain trust is relatively good. The main risks are credential handling and local code execution capability, but there is no sign of unknown egress or clear red flags, so overall it is cautionary rather than high risk.

CredentialsCaution

It requires database connection secrets/parameters (MCP_KEY_HOST, PORT, NAME, USER, PASS, SSLMODE), which are sensitive; if exposed, they could enable unauthorized database access or misuse.

Network egressLow risk

No remote endpoint is stated, and it is described as a local database connection wrapper; there is no evidence of sending user data to unknown third parties.

Code executionCaution

The system check marks executes-code, meaning the tool can run local processes/code or use local system capabilities; this is common for MCP tools, but least privilege and isolation should be considered.

Data accessCaution

The tool targets PostgreSQL data access and is described as having tool selection with safe read-only defaults; however, if write access or a high-privilege DB account is granted, it could still read or modify business data.

Supply chainCaution

It comes from the official Registry with an open-source repository and recent maintenance, which improves supply-chain trust; however, the license is unspecified and the repo has 0 stars, so auditability and adoption evidence are limited.

Recommendations

Use a least-privilege database account, preferably read-only.
Store secrets in a managed secret store; do not hardcode or share config files.
Isolate the runtime so the tool process cannot access unnecessary local resources.
Review the repository and dependencies before deployment, especially write paths and auth handling.

Audited by: gpt-5.4 · 2026-06-17

// install

Copy the install command and let the AI configure it · recommended for beginners

Please install the "Postgres MCP" MCP server from askskill:
Run: claude mcp add 'io-github-edelciomolina-postgres-mcp' -- npx -y @edelciomolina/postgres-mcp