database-migrations-sql-migrations
Pass
Audited by Gen Agent Trust Hub on Aug 8, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill uses a
$ARGUMENTSplaceholder inresources/implementation-playbook.mdto include user requirements without boundary markers or instructions to ignore embedded commands. This creates a surface where a user could potentially override the agent's safe migration logic. - Ingestion points:
$ARGUMENTSvariable inresources/implementation-playbook.md. - Boundary markers: No delimiters or safety instructions are used around the interpolated requirements.
- Capability inventory: The skill provides templates for executing arbitrary SQL via Python (
psycopg2,alembic) and Bash scripts. - Sanitization: No input validation or escaping is applied to the requirements before they are processed by the agent.- [DYNAMIC_EXECUTION]: Several code templates in
resources/implementation-playbook.mduse unsafe string interpolation for database identifiers, creating a potential SQL injection surface in the code the agent might generate. - Evidence: The use of f-strings to inject table names directly into SQL statements, such as
f"SELECT COUNT(*) FROM {table['name']}"invalidate_post_migrationandf"INSERT INTO v2_{table_name}"inParallelMigrator. If the agent populates these placeholders with untrusted user input, it could lead to unauthorized database operations.
Audit Metadata