The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes several scripts to automate presentation handling and validation.
scripts/export_ppt_render.ps1 and scripts/merge_verified_pages.py use the PowerPoint COM (Component Object Model) interface to programmatically control the Microsoft PowerPoint application for exporting slides and merging files.
scripts/test_validate_pptx.py uses subprocess.run to execute Python validation scripts as part of its internal unit testing suite.
[PROMPT_INJECTION]: The skill is subject to indirect prompt injection risks as it is designed to ingest and analyze untrusted user-supplied documents in various formats (DOCX, PDF, TXT, XLSX).
Ingestion points: Source materials uploaded by the user are processed during the 'Analysis' phase as defined in SKILL.md and references/source-analysis.md.
Boundary markers: The skill does not explicitly use delimiters like XML tags or triple quotes to wrap user input in its interpolation instructions.
Capability inventory: The skill possesses capabilities for file system operations (reading/writing JSON and PPTX), XML manipulation, and controlling local applications through PowerShell and Python COM libraries.
Sanitization: No explicit sanitization or filtering of input data is documented, though the skill implements an 'Evidence Gate' which requires all AI-generated facts to be traceable to source locations, serving as a logical validation layer.

cyber-ppt