The Agent Skills Directory

[SAFE]: No security issues detected. The skill performs local image processing and PowerPoint parsing using established libraries. All network operations are limited to standard package installations or font fetching from well-known services (Google Fonts, Fontshare).\n- [EXTERNAL_DOWNLOADS]: The skill installs python-pptx and Pillow via pip. These are widely used libraries for PowerPoint manipulation and image processing. The skill also references fonts from Google Fonts and Fontshare, which are well-known, trusted services for web typography.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided text and external .pptx files to generate slide content, which constitutes an indirect prompt injection surface.\n
Ingestion points: Reads .pptx files and user-provided notes in SKILL.md (found in Phase 1 and Phase 4).\n
Boundary markers: None explicitly defined in the provided instructions; the agent relies on its own inherent instruction boundaries.\n
Capability inventory: Uses local file reading/writing and execution of Python transformation scripts.\n
Sanitization: No explicit sanitization or filtering of external content is described, although the results are contained within isolated HTML/CSS files.\n- [COMMAND_EXECUTION]: Local Python scripts are utilized for slide data extraction and image editing (cropping/resizing). The code snippets provided in SKILL.md are transparent, follow least-privilege principles by operating only on specified input/output paths, and are restricted to their stated utility purposes.

openclaw-slides