The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied project descriptions and entity requirements to generate ORM models, project configurations, and deployment scripts. This creates an attack surface where a malicious user could attempt to embed instructions within their project description to influence the generated code or commands.
Ingestion points: User prompts requesting the creation of new Tortoise models, projects, or migrations as outlined in SKILL.md.
Boundary markers: The skill does not provide explicit instructions or delimiters to help the agent distinguish between user data and instructions.
Capability inventory: The skill guides the agent to write various Python files (models/*.py, config/tortoise.py) and generate shell commands for database migrations (aerich upgrade) within docker-compose.yml, Helm charts, and other deployment manifests.
Sanitization: No explicit sanitization or validation logic is specified for user-provided identifiers or requirements before they are interpolated into code or scripts.
[COMMAND_EXECUTION]: The skill provides templates for automated database migration execution in production-like environments. It specifically instructs the agent to propose adding uv run aerich upgrade commands to docker-compose.yml services, Helm pre-install/pre-upgrade Jobs, and Heroku Procfile release phases. While these are standard practices for managing database state, they involve shell command execution derived from the skill's logic.

tortoise-dev