dba
Warn
Audited by Gen Agent Trust Hub on Jul 25, 2026
Risk Level: MEDIUMDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The skill instructions direct the agent to access sensitive configuration files to establish database connectivity. \n
- Evidence: In
references/discover.md(Step D3), the agent is instructed to search forDATABASE_URLor similar patterns within.env*files to determine the database host. \n - Mitigation: The skill includes a specific safety instruction to read only the variable name and host and not to print or expose actual credentials.\n- [PROMPT_INJECTION]: The skill's project discovery and diagnostic workflows create an attack surface for indirect prompt injection. \n
- Ingestion points: The discovery protocol (
references/discover.md) reads project documentation (CLAUDE.md, incident reports) and existing migration files. The query library (references/queries.md) processes live database catalog and statistics output. \n - Boundary markers: There are no explicit delimiters or instructions provided to the agent to treat data from these external sources as potentially untrusted. \n
- Capability inventory: The agent has extensive permissions, including shell command execution (
ls,grep), execution of administrative SQL queries, and the ability to write new migration and test files. \n - Sanitization: The skill does not specify any sanitization or validation of the data extracted from external files before it is used by the agent.
Audit Metadata