The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill's rendering script (scripts/render_all.sh) includes a command to install the python-docx library (version 1.1.2) from a standard package registry. This library is a well-known, widely used tool for generating Word documents and is necessary for the skill's primary function of creating editable lesson plans.
[INDIRECT_PROMPT_INJECTION]: The skill contains an attack surface for indirect prompt injection as it is designed to ingest and process external content from user-provided files and URLs in Step 1.
Ingestion points: Lessons are ingested via file uploads or URL fetches in SKILL.md and the subject-specific reference files.
Boundary markers: The instructions do not explicitly mandate that the agent wrap external content in delimiters to prevent the LLM from obeying instructions embedded within the source lessons.
Capability inventory: The skill has the capability to write local files and execute bundled Python scripts to render documents.
Sanitization: The skill mitigates this risk by requiring the agent to transform the data into a rigid, structured differentiation.json schema and follow strict pedagogical rules (R1–R8) before the rendering script processes the content.
[COMMAND_EXECUTION]: The skill uses a bash script (scripts/render_all.sh) to invoke Python-based document renderers. This is an expected operational move for the skill's documented purpose of generating classroom-ready materials.

k12-lesson-differentiation