M
MCP Directory CURATED DIRECTORY
911 MCP servers indexed
中文
← Back to directory
COMPARE UP TO 4 SERVERS

Compare MCP servers

Compare scores, permissions, risks, and fit in one decision-focused table.

DimensionSLayerAn embeddable, expressive semantic layer for AI agents and humansMCP Toolbox for DatabasesGoogle's official database MCP toolbox — define AI-accessible database tools via config
FMRS48 / 100 · D74 / 100 · B
Reliability7 / 2013 / 20
Security and permissions9 / 2013 / 20
Maintenance10 / 2017 / 20
Documentation10 / 2016 / 20
Setup experience12 / 2015 / 20
Best for
  • Data teams wanting a governed data access surface for AI agents
  • Developers building analytical MCP servers or APIs on top of a warehouse
  • Teams with existing dbt, Cube, or Ossie semantic configs they want to reuse
  • Engineers embedding a semantic layer into Python apps
  • Teams that want precise control over which database operations an AI can perform, rather than open arbitrary SQL execution
  • Scenarios needing a unified MCP setup across multiple database engines
Not for
  • Scenarios requiring write access to the database (SLayer connections are read-only)
  • Simple cases where direct SQL is sufficient and no semantic layer is needed
  • Non-Python stacks that cannot run the CLI or an MCP subprocess
  • Lightweight cases that just want to run a few ad-hoc SQL queries without maintaining a tools.yaml config (a simpler single-database MCP may be a better fit)
Required permissions
  • Read-only database access to configured datasources
  • Environment variables holding database credentials such as DB_PASSWORD
  • Permission to run the local CLI and the MCP subprocess
  • Permission to modify Claude Code MCP configuration
  • Database credentials (username/password/connection string) are supplied via env vars or config
  • A tool's actual permission is whatever SQL statement is defined in tools.yaml — designed for least privilege, but misconfiguration can still over-expose access
Risks and side effects
  • Database credentials are passed to the MCP subprocess via environment variables, so a compromised launching shell could expose them
  • Connections are read-only but can still read sensitive business data, so row-level security should be applied
  • Once connected via MCP, agents can run queries, so verify their query scope matches expectations
  • As a third-party project, availability and behavior depend on upstream maintenance
  • If tools.yaml defines SQL statements that allow unconstrained writes or deletes, the AI could accidentally modify data
  • The prebuilt toolsets (--prebuilt) favor convenience and may expose broader query capability than a specific business actually needs — use a custom tools.yaml in production
Supported clientsClaude CodeClaude Code, Gemini CLI, Zed, Antigravity
Tools00