Find where your code builds SQL by concatenation, interpolation or formatting, with the parameterised form to write instead. It cannot prove a query is safe.
Waiting for code
Paste the code that builds your query. Nothing is executed and nothing is uploaded.
Waiting for code
Paste the code that builds your query — JavaScript, Python, PHP, Java or C#.
?MySQL, SQLite, JDBC, and most ODBC drivers.$1PostgreSQL, and node-postgres.:nameOracle, SQLAlchemy, better-sqlite3 and many ORMs.@nameSQL Server and the .NET drivers.%spsycopg and MySQLdb. Not a format string, despite how it looks.A placeholder stands in for a value, never for a table or column name. A dynamic ORDER BY has to be checked against a list you control — escaping an identifier is not the answer.
SQL Injection Inspector runs entirely on your device — nothing is uploaded, and no code is executed.
You’re on the Free plan. Compare plans
Static inspection cannot prove a query is safe, and this tool says so rather than issuing a verdict. It finds query strings built by concatenation, interpolation or formatting — the shapes that make injection possible — and it cannot see whether a value came from a user, whether something upstream validated it, or whether an ORM is parameterising underneath. A clean result means nothing matched the patterns.
Escaping is not parameterisation, and conflating them is the mistake behind most injection that survives a code review. Escaping makes one value safe to embed, correctly, today, and relies on the next person remembering. Parameterisation sends the query and the values separately, so the database never parses a value as SQL and there is nothing to remember.
Identifiers cannot be parameterised, and this is the case people get wrong after they have learned to parameterise everything else. A placeholder stands in for a value, never for a table or column name — so a dynamic `ORDER BY` has to be checked against a list you control rather than escaped.
The placeholder syntax differs by driver and the mismatch is a common source of confusion: `?` in MySQL and SQLite, `$1` in PostgreSQL, `:name` in Oracle and many ORMs, `@name` in SQL Server, and `%s` in Python's DB-API — which is not a format string despite looking like one, and using it as one is itself the vulnerability.
This is a defensive tool. It contains no exploit payloads, names the shape of the problem and says what to write instead.
No. It means nothing matched the patterns. This reads shapes in text — it cannot tell whether a value came from a request, whether it was validated upstream, or whether your ORM parameterises underneath. Treat it as one check among several.
It works when done correctly every time, which is the problem. Parameterisation removes the question: the database receives the query and the values separately and never parses a value as SQL. Escaping relies on a person; parameters do not.
Check it against a list of allowed columns that you control. A placeholder cannot stand in for an identifier, and escaping one is not equivalent — the only safe dynamic identifier is one you chose from a fixed set.
Because it depends entirely on how it is used. Passed as the second argument to `execute` it is a placeholder and is safe. Used with `%` or `.format()` to build the string first, it is ordinary string formatting and the parameterisation never happens.
No. It is read in your browser, never executed, and never sent anywhere.