ppt-editing-skill

The ppt-editing-skill is a specialized utility designed for AI agents to perform surgical, template-based updates on existing PowerPoint (PPTX) presentations. Unlike general-purpose tools, this skill treats the PPTX structure as an editable XML tree, allowing agents to manipulate slides, reorder sequences, and update text content with high precision while maintaining file integrity. It is ideal for enterprise use cases where consistent branding and complex layouts must be preserved during content rotation or bulk updates.

• Precise XML manipulation of slide content, including text replacement and layout restructuring.

• Structural control: safely delete, duplicate, or reorder slides via direct modification of p:sldIdLst.

• Parallel editing capability: since each slide is represented by an individual XML file, multi-agent workflows can process and edit slides simultaneously to improve efficiency.

• Automated cleanup: integrated scripts for removing orphaned media, broken relationship IDs, and unused assets to prevent file bloat.

• Validation-first approach: uses pack/unpack utilities to ensure the output PPTX remains compliant with the OpenXML standard and prevents structural corruption.

• Analyze templates using markitdown to extract structure and placeholder text before beginning modifications.

• Use the provided scripts (unpack.py, add_slide.py, clean.py, pack.py) to manage the document lifecycle.

• Formatting enforcement: supports bold header application using b=1 on a:rPr, proper list formatting with a:buChar, and smart quote conversion to ensure professional output.

• Practical constraints: always write outputs to a temporary directory before final saving to circumvent limitations with zipfile operations on specific volume mounts.

• Content mapping: focus on matching input sections to varied template layouts (e.g., multi-column, full-bleed images, icon grids) to avoid monotonous slide design.

• Template adaptation: when replacing placeholders, verify content overflow and remove entire unused shape groups rather than just clearing text to maintain design consistency.