The Agent Skills Directory

[COMMAND_EXECUTION]: The helper script scripts/render_export.py utilizes subprocess.run to call the pdftoppm utility (part of the standard poppler-utils package). This functionality is used to render PDF exports from Google Drive into images, allowing the agent to perform a visual check of the document layout as described in the skill's verification procedures.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process content from external sources (Google Docs, Sheets, and Slides), which presents a surface for indirect prompt injection if those files contain malicious instructions.
Ingestion points: Data enters the agent's context through the Google Drive, Docs, Sheets, and Slides connectors as described in SKILL.md.
Boundary markers: The reference files (references/docs.md, references/sheets.md, references/slides.md) provide extensive writing and formatting rules intended to ensure a professional and consistent output, effectively serving as behavioral guardrails.
Capability inventory: The skill possesses capabilities to execute shell commands (pdftoppm in render_export.py) and perform complex file and data manipulation within the Google Workspace environment.
Sanitization: The helper scripts include logic to unwrap and validate JSON data from tool results and verify PDF file headers before processing.

google-workspace