mysql-mcp-server
Provides an MCP server (and optional HTTP/REST mode) that connects to one or more MySQL/MariaDB instances and exposes read-only database introspection tools and controlled, row-limited SQL querying plus optional vector search. Intended for agent clients like Claude Desktop/Cursor to safely explore schemas and run safe read-only queries.
Score Breakdown
⚙ Agent Friendliness
🔒 Security
Strengths: supports TLS for MySQL connections (including skip-verify/preferred mapping), encourages read-only usage, provides query timeouts and max row caps, and mentions audit logs. Concerns/unknowns: no explicit app-layer authentication/authorization for the optional HTTP/REST mode is described in the provided README; HTTP exposure therefore should be protected via network controls. Scope granularity is limited to tool-level read-only enforcement rather than cryptographically scoped permissions. Secrets appear to be supplied via DSN/env vars; the README does not explicitly state whether secrets are redacted from logs (structured logging exists, but redaction is not documented).
⚡ Reliability
Best When
You can run the server locally or inside a controlled network with a least-privilege DB user and you need agent-friendly read-only MySQL introspection via MCP (optionally with row caps and timeouts).
Avoid When
You cannot restrict network access to the MCP/HTTP endpoints or you must guarantee strict safety beyond read-only SQL filtering and row limits.
Use Cases
- • Schema exploration and documentation (list databases/tables, describe tables)
- • Read-only debugging assistance for MySQL/MariaDB
- • Building internal tooling for safe analytics over operational databases
- • Vector similarity search on MySQL 9.0+ vector columns
- • Connection/pool health and server info retrieval (ping, server_info, connections)
Not For
- • Modifying data (INSERT/UPDATE/DELETE/DDL)
- • Running unbounded or expensive queries without limits
- • Public internet exposure without network controls/auth at the HTTP layer
- • Handling production workloads without careful DSN privilege and resource sizing
Interface
Authentication
No separate app-layer auth is documented for MCP stdio. For optional HTTP mode, auth/authN is not described in the provided README content; HTTP exposure should be treated as needing external network protection.
Pricing
Self-hosted open-source style package; no pricing info in provided content.
Agent Metadata
Known Gotchas
- ⚠ Safety relies on SQL type restrictions (SELECT/SHOW/DESCRIBE/EXPLAIN) and row limits; agents may still craft expensive SELECTs unless timeouts/row caps are configured.
- ⚠ HTTP mode is optional and may be mutually exclusive with stdio MCP; ensure correct env vars (MYSQL_MCP_HTTP vs stdio) to avoid unexpected interface behavior.
- ⚠ Token dashboard availability depends on token tracking and the metrics HTTP sidecar settings (especially for stdio mode).
- ⚠ Multi-DSN tool behavior depends on correct configuration of MYSQL_DSN_N and *_NAME/ *_DESC fields.
Alternatives
Full Evaluation Report
Comprehensive deep-dive: security analysis, reliability audit, agent experience review, cost modeling, competitive positioning, and improvement roadmap for mysql-mcp-server.
AI-powered analysis · PDF + markdown · Delivered within 30 minutes
Package Brief
Quick verdict, integration guide, cost projections, gotchas with workarounds, and alternatives comparison.
Delivered within 10 minutes
Score Monitoring
Get alerted when this package's AF, security, or reliability scores change significantly. Stay ahead of regressions.
Continuous monitoring
Scores are editorial opinions as of 2026-03-30.