Skip to main content
Glama
surinch-db-guard

surinch-db-guard

@surinch/db-guard-mcp

Local-first, zero-dependency deterministic DDL governance scorer, SQL readonly safety reviewer, and metadata-only sensitive masking scanner (CLI + MCP Server) with SHA-256 verifiable receipts.

License Node.js Zero Dependencies

@surinch/db-guard-mcp gives AI coding agents (Cursor, Claude Desktop, Windsurf, Antigravity, DSH) and CI pipelines a deterministic, zero-hallucination database safety boundary. Everything runs locally in Node.js with zero external network calls, zero database write privileges, and cryptographically reproducible surinch.receipt.h1 SHA-256 audit receipts.


Core Tools (MCP & CLI)

MCP Tool

CLI Subcommand

What It Does

Hard Safety Guarantees

ddl_governance_check

ddl-check

Scores CREATE TABLE DDL (mysql, postgresql, doris) across 5 weighted dimensions (25% naming, 20% comments, 20% primary key, 20% audit monitoring, 15% type precision). Supports real pg_dump (PG 14–17) and mysqldump (MySQL 8.0/8.4).

Flags missing PKs and FLOAT/DOUBLE monetary columns as P0. Rejects empty input, comment-only SQL, and unsupported dialects with explicit error codes (never fake green).

sql_readonly_review

sql-review

Static readonly safety gate & anti-pattern reviewer (mysql, postgresql). Lexically strips comments and string literals first so REPLACE(...) and 'DROP' string literals never trigger false positives.

Blocks EXPLAIN ANALYZE, write/DDL statements, CTE writes, and DELETE/UPDATE without WHERE as P0. Flags SELECT *, NOT IN (SELECT ...), leading wildcards, WHERE column functions, deep OFFSET, and PG integer division.

sensitive_masking_scan

mask-scan

Metadata-only sensitive column classifier (L4 credentials/secrets, L3 national ID / bank / phone, L2 email / name / address / IP) and deterministic CREATE OR REPLACE VIEW masking SQL generator (mysql, postgresql, doris).

Never touches sample row data—inspects schema metadata only. Redacts L4 credentials to NULL by default and validates viewPrefix against SQL injection.

Verifiable Receipt (surinch.receipt.h1)

Every MCP call and CLI --receipt invocation emits a deterministic receipt:

{
  "schema_version": "surinch.receipt.h1",
  "tool": "ddl_governance_check",
  "engine_version": "0.1.0",
  "verdict": "PASS",
  "input_sha256": "<sha256 of raw input bytes>",
  "output_sha256": "<sha256 of canonical sorted-key JSON result>",
  "generated_at": "2026-11-19T12:00:00.000Z"
}

Given the same input and engine_version, input_sha256 and output_sha256 are byte-for-byte identical across machines and can be independently verified offline.


Related MCP server: PreSQL

One-Click MCP Client Configurations

Prerequisites: Node.js >= 20.0.0. No API keys, no database credentials, and no background daemon required.

1. Cursor (~/.cursor/mcp.json or .cursor/mcp.json)

{
  "mcpServers": {
    "surinch-db-guard": {
      "command": "npx",
      "args": ["-y", "@surinch/db-guard-mcp"]
    }
  }
}

2. Claude Desktop (claude_desktop_config.json)

  • macOS: ~/Library/Application Support/Claude/claude_desktop_config.json

  • Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "surinch-db-guard": {
      "command": "npx",
      "args": ["-y", "@surinch/db-guard-mcp"]
    }
  }
}

3. Windsurf (~/.codeium/windsurf/mcp_config.json)

{
  "mcpServers": {
    "surinch-db-guard": {
      "command": "npx",
      "args": ["-y", "@surinch/db-guard-mcp"]
    }
  }
}

4. Antigravity (mcp_config.json)

{
  "mcpServers": {
    "surinch-db-guard": {
      "command": "npx",
      "args": ["-y", "@surinch/db-guard-mcp"]
    }
  }
}

5. DSH / Local Global Install

If you prefer offline execution without npx:

npm install -g @surinch/db-guard-mcp

Then configure your MCP client (DSH or any stdio MCP host):

{
  "mcpServers": {
    "surinch-db-guard": {
      "command": "surinch-db-guard-mcp"
    }
  }
}

CLI Usage (CI/CD & Local Terminal)

# 1. DDL governance check (exits 0=PASS/WARN, 1=P0 BLOCK, 2=INVALID_INPUT)
cat schema.sql | npx -y -p @surinch/db-guard-mcp surinch-db-guard ddl-check --dialect postgresql --receipt

# 2. SQL readonly safety review
cat query.sql | npx -y -p @surinch/db-guard-mcp surinch-db-guard sql-review --dialect mysql --receipt

# 3. Metadata-only sensitive column scan & masking VIEW SQL generation
npx -y -p @surinch/db-guard-mcp surinch-db-guard mask-scan \
  --dialect postgresql \
  --view-prefix v_masked_ \
  --file schema.sql \
  --receipt

Programmatic ESM Usage

import {
  analyzeDdlGovernance,
  reviewSqlReadonly,
  scanSensitiveMasking,
  buildReceipt,
} from '@surinch/db-guard-mcp';

const ddl = `CREATE TABLE orders_fact (
  order_id BIGINT PRIMARY KEY COMMENT 'pk',
  pay_amount DECIMAL(18,2) NOT NULL COMMENT 'amount',
  created_at DATETIME NOT NULL COMMENT 'ts'
) COMMENT='orders';`;

const result = analyzeDdlGovernance(ddl, 'mysql');
const receipt = buildReceipt({ tool: 'ddl_governance_check', rawInput: ddl, result });
console.log(result.totalScore, receipt.verdict, receipt.output_sha256);

Security & Zero-Egress Design

  • Readonly by construction: Never opens network sockets or database connections.

  • Metadata only: sensitive_masking_scan operates purely on DDL column names/types and never ingests row values.

  • Zero third-party runtime dependencies: Uses Node.js built-in modules (node:crypto, node:fs, node:readline, node:url) only.

Online Tools & Enterprise Edition

License

Apache-2.0 © Shanghai Surinch Network Technology Co., Ltd. (上海盈尺网络科技有限公司). See LICENSE and NOTICE.

Available Tools

3 tools
ddl_governance_checkB

Score CREATE TABLE DDL (MySQL/PostgreSQL/Doris) across 5 weighted governance dimensions with SHA-256 verifiable receipt. Rejects empty/invalid input or unsupported dialect.

Inputs: ddl (required), dialect.

sensitive_masking_scanA

Metadata-only sensitive column classifier (L4/L3/L2) & deterministic masking VIEW SQL generator for MySQL/PostgreSQL/Doris DDL with SHA-256 verifiable receipt. Never reads row data.

Inputs: ddl (required), dialect, viewPrefix.

sql_readonly_reviewB

Static readonly safety gate & anti-pattern reviewer for MySQL/PostgreSQL SQL with SHA-256 verifiable receipt. Does not falsely flag REPLACE() or comments.

Inputs: sql (required), dialect.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A3.7/5.0

    Scored across 3 tools

    Disambiguation5/5

    Each tool targets a distinct artifact and action: DDL governance scoring, readonly SQL safety review, and sensitive-column masking. Shared receipt behavior does not blur their purposes, so an agent can select reliably.

    Naming Consistency4/5

    All names are snake_case and follow a readable domain/qualifier/action structure, though the action words vary (check, review, scan). The convention is predictable even if not a strict verb_noun pattern.

    Tool Count5/5

    Three focused tools match the narrow database-guard scope, with each covering a separate governance gate. The surface is minimal without appearing redundant or under-specified.

    Completeness4/5

    Core guard workflows are covered: DDL scoring, readonly SQL review, and sensitive-data masking. Minor gaps exist around receipt verification, policy/config management, or broader dialect coverage, but they do not block the primary use cases.

    Maintenance

    ActivityMaintained
    ResponsivenessNo recent community issues

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      Not graded
      quality
      B
      maintenance
      Gives AI coding assistants, IDEs, and CI full PostgreSQL schema intelligence from an offline snapshot, enabling linting, query validation, migration safety analysis, and foreign key graph exploration without ever exposing database credentials.
      36
      BSD 2-Clause "Simplified"
    • A
      license
      A
      quality
      B
      maintenance
      Enables agents to validate, explain, and gate SQL statements against a supplied schema before execution, returning safe/review/blocked verdicts, issue codes, and fix suggestions. It acts as a pre-execution go/no-go guard and composes with database access servers.
      3
      MIT
    • A
      license
      Not graded
      quality
      B
      maintenance
      Enables AI coding assistants to audit datasets against DAMA-DMBOK quality dimensions, sanitize PII, generate differential-privacy synthetic twins, compress prompt context, and export dbt schema tests and interactive HTML reports. It supports development against sensitive data with zero real records exposed to external LLMs.
      1
      MIT
    • A
      license
      B
      quality
      C
      maintenance
      Enables AI coding agents to perform deterministic database architecture audits and performance engineering for PostgreSQL and MySQL, covering migration analysis with schema health scoring, ESR-based composite index recommendations, N+1 query detection, EXPLAIN plan diagnosis, transaction and deadlock analysis, partitioning and archival design, and capacity or configuration tuning. It also generates phased Expand & Contract runbooks that turn blocking DDL into zero-downtime production migrations.
      8
      MIT