- Home
- Skills
- Data & Databases
- Enterprise-SQL-MCP-Guard
More screenshots
Works with the AI Agents you already use
Enterprise-SQL-MCP-Guard
Hardened read-only SQL agent skill for safe database discovery, PII-redacted analytics, and schema auditing.
$14
Enterprise-SQL-MCP-Guard
Example session with this skill installed
Analyze the billing_ledgers table to find the total overdue_usd grouped by payment_status. verify the schema first.
- Read your context and instructions
- Compiled the enterprise-sql-mcp-guard
I have inspected the database schema and executed a read-only aggregation on the billing ledgers.
Overdue Billing Summary
| payment_status | invoice_count | total_overdue_usd | avg_invoice_value |
|---|---|---|---|
| OVERDUE | 41 | $3,214,500.80 | $78,402.45 |
| DISPUTED | 15 | $1,102,400.15 | $73,493.34 |
| PENDING_WIRE | 82 | $0.00 | $45,210.12 |
| SETTLED | 312 | $0.00 | $42,100.50 |
Telemetry Data:
- Execution Time: 1.42 ms
- Security Status: PASSED
- Truncated: False
Next steps
- Run a join with
institutional_clientsto see which tiers own these overdue invoices. - Filter for specific
client_idvalues to drill down into high-value disputed cases. - Check the
idx_billing_statusindex usage if query performance drops on larger datasets.
Connects securely to your tools. The creator never sees your data.
About this skill
Querying production databases often carries the risk of accidental mutations, performance degradation, or PII exposure. This skill provides a hardened, zero-trust gateway for agents to perform read-only SQL analysis and schema discovery without risking data integrity.
What it does
- Read-Only Enforcement uses a two-tier defense including AST validation and SQLite URI
mode=roto block all DML and DDL operations. - Schema Discovery provides a dedicated tool to inspect tables, columns, indexes, and row estimates without manual
PRAGMAcalls. - Automated Guardrails enforces row-count ceilings, payload byte limits, and execution timeouts to protect the agent's context window.
- Security Sanitization redacts PII and sensitive tokens like API keys or password hashes from query results.
- Query Optimization allows agents to run
EXPLAIN QUERY PLANto verify index utilization before executing complex joins.
How it works
- Schema Discovery: The agent calls
inspect_schemato map the database structure and avoid identifier hallucinations. - Safe Querying: The agent executes
SELECTorWITHstatements via theexecute_readonly_querytool. - Validation: Every query passes through a regex and AST gate to ensure no mutating keywords are present.
- Telemetry Audit: Use
get_server_telemetryto monitor execution latencies, blocked attempts, and server health.
Frameworks & tools
This skill is built for Model Context Protocol (MCP) environments using Python and aiosqlite. It is compatible with Claude Code, Cursor, and other MCP-enabled IDEs.
Why this beats prompting it yourself
Prompting an agent to be "read-only" is unreliable and prone to injection or accidental UPDATE commands. This skill moves enforcement from the prompt to the runtime, ensuring that even if the agent tries to mutate data, the database driver and AST validator will deterministically block it.
Use cases
- Revenue Analytics: Aggregating billing ledgers and client tiers to find payment exposure.
- Data Debugging: Investigating API usage logs for high-latency endpoints or HTTP error spikes.
- Index Auditing: Verifying that production queries are properly using B-Tree indexes to prevent table scans.
- PII-Safe Reporting: Generating business intelligence reports that automatically mask sensitive user data.
Known limitations
Supports SQLite only. Hard ceiling of 100 rows per result and 64KB payload limit. Queries exceeding 3 seconds are automatically terminated(all three thresholds are configurable via environment variables).
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Data & Databases.
SQLite only. Hard 100-row and 64KB payload limits. 3.0s execution timeout. No support for DDL, DML, or administrative commands.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 5 days ago
- Accounts
Sqlite
- Software
python 3.11+SQLite databaseEnterprise Sql Guard MCP
Git
- Optional
- pytestRich CLI
The FastMCP Python server is bundled inside the ZIP—no external paid APIs required. Set MCP_SQL_GUARD_DB_PATH to your local SQLite file (or run scripts/seed_enterprise_db.py to test with a 1,800-row sample database).
Permissions
Allowed Hosts
File Scopes
Strict read-only file access via SQLite URI (?mode=ro) and PRAGMA query_only = ON. Environment variables are used solely to configure the local database path, query timeout (3.0s), and row/payload bounds. Zero external network calls or disk writes are ever performed.