The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process raw user-provided instructions ('raw dictated instruction') to transform them into structured prompts. This represents an attack surface where malicious instructions could be embedded in the input data.
Ingestion points: The skill accepts a single prompt parameter from the user (SKILL.md).
Boundary markers: The prompt instructions do not explicitly define strict boundary markers or 'ignore embedded instructions' directives when handling the raw input.
Capability inventory: The skill itself does not perform file writes, network operations, or shell commands, but it does instruct the agent to output a 'copy-pasteable prompt' which could influence downstream agent actions if the resulting prompt is executed.
Sanitization: No explicit sanitization or filtering of the raw prompt is mentioned; however, the 'Ambiguity Gate' and the 'Rewrite Rules' act as a form of human-in-the-loop and logic-based filtering that reduces the risk of accidental obedience to malicious instructions within the raw input.

promptify