The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses local Python scripts (scripts/init_deck.py and scripts/validate_deck.py) to manage the lifecycle of the generated HTML files. These scripts handle file creation, path slugification, and automated structure validation. Additionally, the skill utilizes shell commands like npx postplan upload to handle the publishing of generated decks.
[EXTERNAL_DOWNLOADS]: The publishing workflow involves npx postplan upload, which fetches and executes the postplan package from a remote registry. This is an intended part of the handoff to the html-communication system and is protected by a mandatory user confirmation prompt before execution.
[INDIRECT_PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection as it ingests untrusted user-supplied topics and content to generate HTML.
Ingestion points: User-provided topic descriptions and source materials (Workflow Step 1 in SKILL.md).
Boundary markers: The process uses a structured blueprint (assets/story-blueprint-template.md) and a rubric (evals/RUBRIC.md) to guide generation.
Capability inventory: The skill can perform local file writes and execute npx commands.
Sanitization: While the skill follows a rigid narrative spine to ensure quality, it does not explicitly define sanitization routines for user strings interpolated into the {{TITLE}} placeholder in assets/base-shell.html.

interactive-teaching-page