The Agent Skills Directory

[PROMPT_INJECTION]: The skill is designed to ingest untrusted article content, reports, or notes provided by the user to generate slide decks. This creates a surface for indirect prompt injection.
Ingestion points: User-provided raw article content or notes processed by references/article_intake_workflow.md.
Boundary markers: The skill does not define explicit delimiters or instructions to the agent to disregard instructions embedded within the processed article content.
Capability inventory: The skill generates several markdown planning artifacts and invokes image generation tools based on the ingested content.
Sanitization: No explicit sanitization or filtering of the input text is described; the skill relies on the underlying model's safety guardrails.
[COMMAND_EXECUTION]: The skill includes several Python scripts for local automation that perform standard file and directory operations within the workspace.
scripts/build_images_only_pptx.py: Assembles generated images into a PowerPoint file using the python-pptx library.
scripts/compose_style_pack_blocks.py: Aggregates style-pack configuration files. It uses ast.literal_eval to safely parse literal values from manifest files.
scripts/create_style_pack.py: Scaffolds new directory structures and templates for style packs. These scripts do not execute arbitrary shell commands or untrusted external code.
[EXTERNAL_DOWNLOADS]: The skill requires the python-pptx library, which users are instructed to install via a standard package manager. This is a well-known and trusted library for PowerPoint automation.

infographic-powerpoint-deck