sql-safe-mcp

A PII-safe, read-only MCP server for coding agents working with SQL Server, MySQL and MariaDB: schema knowledge and safe queries without write-capable tools or plaintext exposure of configured PII columns.

Read the data. Protect the identity.

Give a coding agent the schema knowledge it needs to write correct code - servers, databases, tables, columns, keys, indexes, and stored procedures. On aliases you mark pii_safe, it can also query real rows while configured personal data stays hidden behind authenticated tokens.

PII-safe by default. Read-only by design.

• PII-safe by default - on a pii_safe server, the columns you configure come back as alias-bound, authenticated tokens (pii:v1:...), never as plaintext. An agent can still project, count, and filter on them with = and IN using tokens it was given, so it can follow a record without reading it. Tokens do not work on another server alias or with another key.
• Read-only by construction - seven tools, all annotated read-only. No tool writes data, and the server never executes SQL an agent wrote: metadata comes from SQLAlchemy Inspector and fixed catalog queries, and execute_sql runs only a validated, regenerated SELECT.
• Multiple servers, isolated policies - one MCP process can expose multiple named SQL Server, MySQL, and MariaDB instances. Each alias has its own connection, access level, and, when enabled, PII key and protection rules. Every database operation targets an explicit alias; there is no network discovery.

Operational safeguards

• Fails closed - SQL validation is an allowlist. Unknown syntax, unresolved lineage, and unsupported protected-value types are refused, not guessed at. The verification evidence is in the security model.
• Least access first - access_level: metadata (the default) exposes schema only; execute_sql needs an explicit pii_safe alias with its own key. Database permissions stay the primary control, so use a least-privilege login.
• Secrets stay out of sight - connection URLs live in YAML with ${NAME} placeholders resolved from the environment. They never appear in logs or model-visible errors.
• Compact, predictable output - object-rooted results with stable sorting, literal case-insensitive name filters, and stored procedure lists that do not expand definitions.
• Errors an agent can act on - an ambiguous name lists the candidate schemas. Errors never contain connection details, credentials, keys, tokens, or rows.
• Tested against attacks, not just examples - the SQL and PII boundary is checked with an adversarial corpus of hostile statements, property-based tests (token isolation, tamper resistance), a live attack run against a really writable login with before/after snapshots, and mutation testing. Results are in Verification of the SQL boundary.
• Diagnosable failures - a timeout or connection error carries a Reference, and the stderr log records the connection stage, elapsed time, and driver error for the same reference, with credentials removed. See Logging.

Quick start

Install

uvx sql-safe-mcp

or

pip install sql-safe-mcp

Pin a version when you want a fixed surface: uvx sql-safe-mcp==1.4.0.

Requires Python 3.12+, uv (or pip), and Microsoft ODBC Driver 18 for SQL Server when you connect to SQL Server. MySQL and MariaDB use the bundled PyMySQL driver and need nothing else.

Verified against SQL Server 2022, MySQL 8.4, and MariaDB 11.4 (see CHECKS.md).

Configure multiple servers

Copy sql-safe-mcp.example.yaml to sql-safe-mcp.yaml. Add each database server under a named alias and keep credentials in environment variables:

version: 1
servers:
  reporting:
    engine: sqlserver
    access_level: metadata
    connection_url: "${REPORTING_SQL_URL}"

  billing:
    engine: sqlserver
    access_level: pii_safe
    connection_url: "${BILLING_SQL_URL}"
    pii_key_env: BILLING_PII_KEY
    pii:
      rules:
        - database: Billing
          schema: dbo
          table: Customers
          columns: [Email, FullName]

  shop:
    engine: mysql
    access_level: metadata
    connection_url: "mysql+pymysql://${SHOP_USER}:${SHOP_PASSWORD}@db.internal/shop"

Here reporting, billing, and shop are the values an agent passes as server. Their credentials and access policies are independent. A missing variable, or an engine that does not match the URL dialect, stops the MCP server at startup.

Ask an agent to install it

Give a coding agent the following prompt when you want it to perform the setup in its own MCP host:

Install and configure sql-safe-mcp as a stdio MCP server for this workspace. First inspect the
host's existing MCP configuration convention and preserve all unrelated server entries. Create or
update a non-secret YAML configuration using environment-variable placeholders only. Never put
connection URLs, passwords, PII keys, tokens, bind values, or query results in tracked files, chat
output, or logs. If the database aliases, engines, access levels, PII rules, configuration path,
or secret-storage mechanism are not specified, ask me before choosing them. Configure the host to
run `uvx sql-safe-mcp` with `SQL_SAFE_MCP_CONFIG` and the required variables supplied through the
host's secret or environment mechanism. Run `sql-safe-mcp --check-config` after the variables are
available, then report only the configured aliases and the validation result.

Check the configuration

After defining every environment variable referenced by the YAML file, point the server at it with SQL_SAFE_MCP_CONFIG (or --config) and check it without connecting to any database:

SQL_SAFE_MCP_CONFIG=sql-safe-mcp.yaml uvx sql-safe-mcp --check-config

Configuration is validated at startup, and an error names the problem without printing a URL or secret. Keep credentials in the host's own configuration and never commit them. The server acts with the database account's permissions, so use a dedicated login with the least access the job needs.

Connect an MCP client

Every client configuration needs SQL_SAFE_MCP_CONFIG plus the environment variables referenced by your YAML file. Keep connection URLs and PII keys in the MCP host's environment or configuration, never in the YAML file or other tracked files.

claude mcp add --env SQL_SAFE_MCP_CONFIG=/path/to/sql-safe-mcp.yaml --env REPORTING_SQL_URL=mssql+pyodbc://... --transport stdio sql-safe -- uvx sql-safe-mcp

{
  "mcpServers": {
    "sql-safe": {
      "command": "uvx",
      "args": ["sql-safe-mcp"],
      "env": {
        "SQL_SAFE_MCP_CONFIG": "/path/to/sql-safe-mcp.yaml",
        "REPORTING_SQL_URL": "mssql+pyodbc://..."
      }
    }
  }
}

codex mcp add sql-safe --env SQL_SAFE_MCP_CONFIG=/path/to/sql-safe-mcp.yaml --env REPORTING_SQL_URL=mssql+pyodbc://... -- uvx sql-safe-mcp

PII-safe queries

execute_sql on the billing alias allows one restricted SELECT. For example, these arguments:

{
  "server": "billing",
  "database": "Billing",
  "sql": "SELECT CustomerId, Email FROM dbo.Customers"
}

can return a row such as [42, "pii:v1:..."]: the agent can follow the customer without reading the email address. A token is accepted only in = and IN predicates on the same protected column and server alias. Protection covers the columns you list, so list every column that holds personal data.

Documentation

• Configuration reference - aliases, connection URLs, PII rules, runtime limits, and logging.
• Tool reference - arguments, responses, errors, and the accepted SQL subset.
• Security model - trust boundaries, guarantees, limitations, and verification.
• Checks - local, integration, and security verification commands.

Security

The MCP caller, SQL input, database metadata, rows, and tokens are untrusted. Database permissions remain the primary authorization control - this server never widens them. Read the full security model, including its limitations and verification evidence. To report a vulnerability, use the private channel in SECURITY.md.

Contributing

Setup, checks, the test commands, the branch and commit conventions, and the release model are in CONTRIBUTING.md. Changes that affect someone running the server are recorded in CHANGELOG.md.

License

MIT.