SQL is a family of related dialects, not one completely interchangeable grammar. The selected profile changes token recognition and parse decisions before indentation begins.
Small syntax differences matter
PostgreSQL uses dollar-quoted strings and casts, MySQL commonly uses backtick identifiers, T-SQL has bracketed identifiers and TOP, BigQuery has its own types and clauses, and Snowflake adds warehouse-specific constructs. A generic profile can mistake one of these tokens for punctuation or an ordinary word. Choose the engine that will receive the script, not the engine whose name looks closest. If a script combines templating or several dialects, format only a reviewed supported fragment or keep the original.
Treat parse errors as information
A formatter error can mean malformed SQL, the wrong dialect, an unsupported procedure, a custom delimiter, a template marker, or simply a feature newer than the bundled grammar. SQLClear shows a bounded engine message but does not guess another dialect or repair the source. Re-check the selected profile and reduce the input to the smallest reproducible fragment. The actual database parser remains authoritative for what its installed version accepts.
Review before you run or publish
Keep the original beside the formatted copy and compare literals, quoted identifiers, comments, placeholders, operators, statement boundaries, and vendor-specific clauses. Confirm that the selected dialect matches the actual database and that template markers or procedural blocks are supported. Then use the normal database parser, tests, review, permissions, transaction plan, and backup procedure for the intended change. SQLClear only makes text easier to inspect; it does not execute the script or prove that its behavior, performance, permissions, or data effects are acceptable.