The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user data such as the presentation topic, audience, and tone as defined in Step 1 of SKILL.md. This data is directly interpolated into the narrative structure and the generated Python script without boundary markers (e.g., delimiters) or instructions to disregard embedded commands. Since the generated script has the capability to write to the file system (via prs.save), there is a risk that adversarial input could trick the agent into including malicious Python code (such as file reads or system calls) in the output.
[DYNAMIC_EXECUTION]: The primary purpose of the skill is the generation of complete, runnable Python scripts. As specified in the README.md 'Rules' and SKILL.md 'How It Works', the agent is instructed to provide self-contained, error-free code blocks. Because this code is generated dynamically from user-supplied briefs, it requires human verification to ensure that the logic remains within the intended scope of creating a PowerPoint file and does not perform unintended system operations.

python-pptx-generator