The Agent Skills Directory

[SAFE]: The skill consists entirely of a prompt template and instructional text. It does not include any scripts, executable files, or third-party dependencies.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external data (such as product requirement documents or launch plans) via the $ARGUMENTS placeholder.
Ingestion points: Data enters via $ARGUMENTS in SKILL.md.
Boundary markers: The prompt uses <context> tags to wrap external data, which helps the agent differentiate between instructions and user-provided content.
Capability inventory: No capabilities for file writing, network operations, or subprocess execution are present in the skill.
Sanitization: Standard boundary markers are present, though no explicit filtering is defined.

pre-mortem