The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes user-provided migration requirements through the $ARGUMENTS variable in SKILL.md without boundary markers or specific instructions to ignore embedded commands. Ingestion points: User requirements are passed via the $ARGUMENTS variable. Boundary markers: No delimiters are used to separate user input from system instructions. Capability inventory: The skill has access to powerful tools including Bash, Write, and Edit, and the ability to execute database commands. Sanitization: The templates provided in resources/implementation-playbook.md demonstrate unsafe interpolation of database identifiers (e.g., f"SELECT COUNT(*) FROM {table['name']}") and shell variables (e.g., pg_dump -d $DATABASE), which could be exploited if an agent follows these patterns using untrusted data.
[COMMAND_EXECUTION]: The skill is configured to use the Bash tool for database management tasks such as backups and migration execution. While these operations are within the skill's stated purpose, the reliance on shell execution combined with the lack of input sanitization in documentation templates creates a risk of command injection if user-supplied data is not properly escaped.

database-migrations-sql-migrations