postgresql-crud
Pass
Audited by Gen Agent Trust Hub on Jun 17, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONREMOTE_CODE_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes
psqlandsshcommands to perform its primary database operations. These commands are constructed using variables that are properly quoted to prevent local shell injection. - [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection (Category 8) as it processes untrusted data from external databases. Ingestion points: Database query output processed in
scripts/postgresql_crud.sh. Boundary markers: Absent. Capability inventory: The skill can perform SELECT, INSERT, UPDATE, DELETE, and raw SQL operations. Sanitization: While SQL inputs are escaped, the resulting data is returned to the agent without specific sanitization or delimiters to prevent instruction injection from database content. - [REMOTE_CODE_EXECUTION]: In
ssh-remotemode, a Bash script is executed on a remote server via SSH. The execution uses a hardcoded heredoc template and standard shell quoting for parameters like the SQL query and environment keys, which is an expected pattern for this tool's functionality.
Audit Metadata