The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes several bundled Python scripts (scripts/compose_style_pack_blocks.py, scripts/create_style_pack.py, and scripts/build_images_only_pptx.py) to automate the scaffolding of design styles and the final assembly of the PowerPoint deck. These scripts perform standard file system operations using the pathlib library.
[DYNAMIC_EXECUTION]: The script scripts/compose_style_pack_blocks.py uses ast.literal_eval to safely parse configuration values from local manifest files. This approach is a security best practice for evaluating data structures from text without the risks of arbitrary code execution associated with eval().
[EXTERNAL_DOWNLOADS]: The skill requires the python-pptx library for creating the presentation files. This is a well-known and standard library for PowerPoint automation in the Python ecosystem.
[INDIRECT_PROMPT_INJECTION]: The skill provides a surface for indirect prompt injection as it processes user-provided articles to generate slides.
Ingestion points: Raw article content, sermon notes, or research reports provided by the user (SKILL.md, references/article_intake_workflow.md).
Boundary markers: The workflow uses a multi-artifact model (slides_message_plan.md, slides_display_plan.md, slides_visual_plan.md) to isolate the speaker's narrative from the rendered slide text, acting as a functional boundary.
Capability inventory: The skill can execute bundled Python scripts, write files to the workspace, and invoke image generation tools.
Sanitization: The skill employs strict "Negative constraints" and "Text fidelity hard constraints" in its prompt templates to ensure that the image generator only renders the exact text derived during the planning phase, effectively filtering out potential injection attempts within the source article content.

infographic-powerpoint-deck