h2oai/sql-sidekick

Guardrails: Evaluate SQL injection via SELECT queries

pramitchoudhary opened this issue · 0 comments

Reference: https://github.com/payloadbox/sql-injection-payload-list

Current thinking, we do this as a 2-factor approach,

  • Part 1: Based on rules, patterns and known info on the common string, interpolation/injection is possible within a SELECT query. #34
  • Part 2: Use LLM to flag malicious patterns.

https://github.com/amoffat/HeimdaLLM