database-migration
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides patterns for complex data transformations that involve ingesting records from external database tables (e.g.,
SELECT id, address_string FROM usersin the Data Transformations section). - Ingestion points: Data is read from database columns and processed in Javascript loops for further updates.
- Capability inventory: The skill utilizes database read/write access and schema modification capabilities.
- Sanitization: The examples correctly utilize parameterized queries (Sequelize
replacements) to mitigate SQL injection, providing a strong security baseline. - Boundary markers: The skill does not explicitly define instructions for the agent to ignore potentially malicious instructions embedded within the database text content.
- [COMMAND_EXECUTION]: The skill documentation references the execution of standard CLI tools for database management.
- Evidence: Use of
npx sequelize-cli,npm run typeorm, andnpx prisma migrateare documented as standard procedures for running migrations.
Audit Metadata