vetkit.llm-app.llm-output-sql
A query generated by a language model is executed as raw SQL. A prompt-injected request can make the model write any statement the database user is allowed to run: reading other users' rows, changing data or dropping tables.
Last updated:
Suite: AI code risks
javascript, typescript
A query generated by a language model is executed as raw SQL. A prompt-injected request can make the model write any statement the database user is allowed to run: reading other users' rows, changing data or dropping tables.
Category: LLM SECURITY · Engine rule severity: ERROR
How to fix it
Prefer letting the model pick from parameterised, pre-written queries. If free-form SQL is the feature, run it on a read-only connection restricted to the tables and rows the user may see (row level security, a dedicated role, statement timeout) and reject anything that is not a single SELECT.
References
Source: packages/engine/rules/llm-app/javascript.yml.
python
A query generated by a language model is executed as raw SQL. A prompt-injected request can make the model write any statement the database user is allowed to run: reading other users' rows, changing data or dropping tables.
Category: LLM SECURITY · Engine rule severity: ERROR
How to fix it
Prefer letting the model pick from parameterised, pre-written queries. If free-form SQL is the feature, run it on a read-only connection restricted to the tables and rows the user may see (row level security, a dedicated role, statement timeout) and reject anything that is not a single SELECT.
References
Source: packages/engine/rules/llm-app/python.yml.
Generated from shipped rule definitions at build time. Matching is a review signal, not proof of exploitability. “generic” denotes text matching, not complete language or framework coverage. Runtime configuration and available tools affect whether a rule runs.