@hyuga/llm-safe-sql
github.comExplore documented capabilities and connections for @hyuga/llm-safe-sql.
What your agent can do
Let an LLM propose UPDATE/DELETE, then actually run it, measure the real before/after diff, and always roll back — so the human approves measured facts, not the model's claim.
Via CLI.
Tasks depend on the interface and permissions.
Connection methods
Choose a route that fits your agent.
CLI
For agents with a terminal
View setup for CLICLI connection
CLI package
@hyuga/llm-safe-sql