The Agent Skills Directory

[PROMPT_INJECTION]: Indirect prompt injection surface detected. The skill is designed to ingest and process untrusted external data which could contain malicious instructions.
Ingestion points: SKILL.md and references/workflow-rulebook.md identify lesson notes, ChatGPT transcripts, and photographed documents as primary inputs.
Boundary markers: Absent. There are no instructions provided to the agent to treat this external content as data only or to ignore embedded instructions.
Capability inventory: The skill utilizes the agy CLI for drafting/marking and Anti-Gravity for image generation, implying execution of shell commands and external API calls.
Sanitization: Absent. The instructions do not specify any validation or sanitization of the input materials before processing.
[COMMAND_EXECUTION]: The skill heavily relies on an external tool called agy CLI for model interaction and workflow management.
The workflow involves passing extracted text from potentially untrusted sources to this CLI.
Without explicit sanitization of the input strings, there is a risk of command injection if the underlying agent implementation does not properly escape metacharacters when invoking the CLI on the host system.

exam-creator-skill