The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill acts as a software engineering assistant, ingesting user-provided code, feature requirements, and error logs across its primary workflows in SKILL.md and the workflows/ directory.
Ingestion points: User-supplied specifications and existing codebases provided during implementation, debugging, and optimization tasks.
Boundary markers: The instructions lack explicit delimiters or warnings to treat processed content as untrusted data.
Capability inventory: The skill utilizes file writing and command-line tool execution (e.g., bun, uv) to perform its duties.
Sanitization: While reference/coding-standards.md provides excellent security guidance for the code the agent produces, the skill itself does not implement sanitization for the inputs it receives from users.
[SAFE]: The skill documents and encourages industry-standard security practices, including SQL injection prevention, secrets management via environment variables, and input validation in reference/coding-standards.md. It uses well-known developer tools and registries without suspicious remote execution patterns.

engineering