MCP database security: your agent checks SQL as it writes it

A remote MCP server, with nothing to install

Vericto runs a remote MCP server (Streamable HTTP) that authenticates with an API key from your workspace with the ci_dryrun:execute scope. A connected agent calls check_sql while it generates SQL and gets the same verdict as the CLI or the API: BLOCKED, FLAGGED, MONITORED or ALLOWED, with the rule that fired and a suggested fix.

claude mcp add --scope user --transport http vericto https://mcp.vericto.com/mcp \
  --header 'Authorization: Bearer ${VERICTO_API_KEY}'

Tested with Claude Code and Kiro. The docs include the configuration for Antigravity, GitHub Copilot in VS Code and any client that supports remote MCP servers over HTTP with headers.

Six tools

• check_sql: evaluates a script or a list of statements against the workspace rules and returns a verdict per statement, with its line.
• whoami: plan, checks left this month, ruleset version and the workspace databases with their dialect.
• list_rules and show_rule: the effective rule catalogue and the detail of each rule, including its AST condition.
• list_reports and show_report: the check history and the full result of each one.

Sensitive columns over MCP

If a finding carries approved_query, the query reads a masked column: that is the only approved query and the agent must run it instead of the original. MCP cannot force it; only the TCP proxy enforces masking.

Quota and security

• Checks per month, shared with the CLI and the validation API: 1,000 on Free, 10,000 on Builder, unlimited on Team and Enterprise. Each evaluated statement counts as one check; the read-only tools use no quota.
• The API key determines the workspace, and a key with only ci_dryrun:execute cannot read the audit trail or change rules.
• Vericto only analyses the SQL text: it does not connect to your database or run anything.

Keep reading

• Docs: AI agents (MCP)
• The CLI on npm: the same verdict in CI
• Use case: CI/CD
• Data masking