sqlsure Review 2026: Catch Silently-Wrong AI-Generated SQL

A query can be perfectly valid, run without error, and return a number that’s silently wrong — revenue double-counted by a join, an average summed, a patient ID exposed. Databases don’t catch it, linters don’t catch it, and an LLM reviewing its own SQL doesn’t catch it. sqlsure does.

What is sqlsure?

sqlsure is a deterministic SQL semantic inspector. You point it at a query and a “rulebook” of facts your team already declared (dbt tests, PK/FK relationships, or a live DB introspection), and it flags semantic errors in about 0.1 ms — before the query runs. It’s built for AI-written SQL in particular, where silently-wrong output is common.

Key features

• Deterministic checks — catches fan-out double-counting, wrong joins, and exposed PII.
• Three doors — a CI gate (exit 1 on violations), an MCP server for agents, and an embeddable library.
• Reuses what you have — dbt tests become grain, relationships become join cardinality, one-line tags mark what’s safe to sum.
• Self-repairing agents — every rejection carries a machine-actionable fix, so an agent can loop draft → check → fix → check → execute.
• Private by default — offline, no data access, no telemetry; it parses query text only.
• Audited — over BIRD + Spider gold answers: 2,568 queries, 45 flags, zero false alarms.

Who should use it?

sqlsure is for anyone shipping AI-written SQL — text-to-SQL products, agent pipelines, or just Claude Code generating reports. Drop it in CI or as an MCP gate so wrong queries never execute. It pairs naturally with dbt-based stacks.

It’s focused on correctness, not performance tuning, and it needs you to declare semantics (even loosely via introspection) before it can judge.

Pros and cons

Pros: catches the bugs linters miss, deterministic and offline, agent self-repair, zero false alarms in its benchmark audit.

Cons: requires declaring semantics; centered on correctness rather than performance; early rulebook (v0.1).

Pricing

Free and open source — install via pip install sqlsure (Apache-2.0).

FAQ

Does sqlsure connect to my database? No. It parses query text only and never accesses data; introspection reads schema metadata (PK/FK), not row data.

Can my agent fix queries automatically? Yes. Each rejection includes a machine-actionable fix; in their benchmark, applying the fix verbatim produced a passing query 10/10 times.

How is it different from a linter? Linters catch syntax and style; sqlsure checks semantics — whether the query computes the right thing given your declared grain and relationships.