The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest user-supplied outlines and source material to generate deck content. This data is interpolated directly into an HTML template (assets/template.html), creating a surface for indirect prompt injection. A malicious outline could contain instructions designed to manipulate the agent's behavior during the preparation, image generation, or validation phases.
Ingestion points: User-provided outline and source material described in SKILL.md (Step 1).
Boundary markers: None identified; the skill does not use specific delimiters or instructions to ignore embedded commands in user data.
Capability inventory: The agent has capabilities to create directories, copy files, write to the file system, and execute a local Node.js script (scripts/validate-deck.mjs).
Sanitization: No explicit sanitization or escaping of user-provided text is mentioned before it is written to the HTML index file.
[COMMAND_EXECUTION]: The skill workflow relies on the execution of shell commands (mkdir, cp) and a Node.js script (node scripts/validate-deck.mjs) to manage the deck's lifecycle. While these are standard development operations, they represent a direct interaction with the host environment's shell driven by instructions in SKILL.md.

tightened-slide