mysql-crud
Pass
Audited by Gen Agent Trust Hub on Jun 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The scripts/mysql_crud.sh script executes local and remote shell commands using mysql, ssh, and bash to facilitate database interaction and management.
- [DATA_EXFILTRATION]: The skill can retrieve database connection strings from remote .env files via SSH. It includes a redaction mechanism (redact_url) to ensure that passwords and sensitive components of the URL are not exposed in logs or agent output.
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection by processing data from external MySQL databases. (1) Ingestion points: SQL query results from the select, schema, and raw-sql commands enter the agent context via scripts/mysql_crud.sh. (2) Boundary markers: The script formats all database output into structured JSON. (3) Capability inventory: The skill can execute SQL queries, manage local profile files, and run commands on remote servers via SSH. (4) Sanitization: Database output is processed through a json_escape function to ensure valid JSON formatting.
Audit Metadata