MCP-Hub: Universal Database, API & Filesystem Gateway

Querying disparate data sources often requires manual schema hunting, context switching between tools, and handling security risks like SQL injection or path traversal. This skill provides a unified interface for AI agents to interact with databases, REST APIs, and filesystems through a secure gateway.

What it does

• Multi-source orchestration manages SQLite, PostgreSQL, REST/OpenAPI, and local filesystems through a single namespaced interface.
• Intelligent schema introspection discovers table definitions, foreign keys, and API endpoints automatically before execution.
• Secure data access enforces read-only defaults, path-traversal protection, and automatic credential masking for sensitive keys.
• Optimized querying utilizes parameterized SQL, automatic caching, and pagination to handle large datasets efficiently.
• Diagnostic monitoring tracks adapter health and execution history through built-in audit logs and health checks.

How it works

1. Discovery phase invokes hub__list_adapters to identify active data sources and describe tools to map schemas or endpoints.
2. Planning phase constructs parameterized SQL or structured REST requests based on the discovered metadata.
3. Execution phase calls the namespaced adapter tool to retrieve data while respecting sandbox and rate limits.
4. Synthesis phase formats the raw JSON or SQL results into readable tables or summaries.

Frameworks & tools

This skill is designed for MCP-Hub gateways and compatible AI coding environments like Claude Code, Cursor, Copilot, and Windsurf. It supports SQLite, PostgreSQL, OpenAPI v3 specifications, and standard filesystem operations.

Why this beats prompting it yourself

Prompting an agent to "query my database" often leads to hallucinated table names or dangerous raw SQL strings. This skill enforces a strict introspection-first workflow and parameterized execution that prevents injection attacks and runtime errors.

Use cases

• Query production SQL databases using natural language without writing manual joins.
• Document and test REST API endpoints by fetching real-time data from OpenAPI specs.
• Search and analyze local files within a secure, sandboxed environment.
• Troubleshoot data connectivity issues using unified health and audit logs.

Known limitations

Does not support non-relational databases beyond REST interfaces. Write operations require manual configuration as the system defaults to read-only for security.