The Agent Skills Directory

[COMMAND_EXECUTION]: The skill frequently uses subprocess.run to execute system binaries including soffice, pdftoppm, and git. Most notably, scripts/office/soffice.py invokes gcc to compile C source code into a shared object file at runtime.- [COMMAND_EXECUTION]: The script scripts/office/soffice.py implements process injection by loading a dynamically compiled shared library via the LD_PRELOAD environment variable. This technique shims the LibreOffice process to bypass environmental restrictions on Unix domain sockets, which is a high-risk behavior usually associated with malware or sandbox escapes.- [EXTERNAL_DOWNLOADS]: The skill documentation requires the manual installation of several third-party dependencies from public registries. This includes Python packages like markitdown and Pillow, and Node.js packages like pptxgenjs, react-icons, and sharp.- [PROMPT_INJECTION]: The skill facilitates the ingestion of untrusted data from .pptx files. This content is converted to text and images and then presented to the agent and subagents for visual inspection. This introduces a surface for indirect prompt injection where instructions hidden in the slide content could manipulate the agent's behavior.
Ingestion points: SKILL.md, editing.md (content extracted via markitdown).
Boundary markers: Absent.
Capability inventory: scripts/office/pack.py (file-write), scripts/office/soffice.py (subprocess execution).
Sanitization: Absent.

pptx