The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides automated patterns for generating database schemas and migrations based on user input. This creates an attack surface where malicious or malformed schema descriptions provided by an untrusted source could influence the agent to generate and execute unintended database operations.
Ingestion points: User-provided database requirements, model definitions, and migration messages used in scripts/migrate.sh.
Boundary markers: Absent. The skill does not provide specific instructions to the agent to ignore embedded commands in user-provided data.
Capability inventory: The skill has the ability to execute shell scripts (scripts/migrate.sh) that trigger Alembic CLI operations and Python scripts (scripts/init_db.py) that modify database structures.
Sanitization: No explicit validation or escaping is applied to user-provided migration messages before they are passed to shell commands.
[DYNAMIC_EXECUTION]: The database initialization script uses dynamic loading to register models at runtime, which could be exploited if the local environment is compromised.
Evidence: The script scripts/init_db.py contains a wildcard import from app.models import * inside the init_database function.
Pattern: This dynamic import pattern is used to register all models with SQLModel.metadata before table creation. While standard for this workflow, it relies on the integrity of the local app/models.py file.

sqlmodel-expert