Summary

extract_image() in extract_content.py writes img.blob bytes from PPTX image parts directly to disk without content validation, size limits, or path sanitization. A crafted PPTX file could deliver oversized payloads or exploit downstream consumers of the extracted images.

Location

• File: .github/skills/experimental/powerpoint/scripts/extract_content.py
• Function: extract_image() — img.blob write to disk

Risk Assessment

• Severity: HIGH
• Attack Vector: A crafted PPTX file containing malicious image blobs (oversized, polyglot, or path-traversal filenames) could abuse the extraction process
• Impact: Disk exhaustion via oversized blobs, path traversal if filenames are derived from PPTX metadata, or delivery of malicious content disguised as images
• CVSS Category: CWE-434 (Unrestricted Upload of File with Dangerous Type) / CWE-22 (Path Traversal)

Current Behavior

# Writes raw blob bytes from PPTX image part directly to output path
with open(output_path, 'wb') as f:
    f.write(img.blob)

The image blob is written without:

• Validating that the content is actually an image (magic bytes / content-type verification)
• Enforcing maximum file size limits
• Sanitizing the output path against directory traversal
• Checking for polyglot files (files valid as both image and executable formats)

Expected Behavior

Image extraction should include defensive checks:

1. Size limit: Reject blobs exceeding a reasonable maximum (e.g., 50 MB)
2. Content validation: Verify magic bytes match expected image formats (PNG, JPEG, EMF, WMF, SVG)
3. Path sanitization: Ensure output paths are confined to the expected output directory (no ../ traversal)
4. Filename sanitization: Strip or reject filenames containing path separators or null bytes

RPI Framework

task-researcher

• Identify all locations where PPTX blob data is written to disk
• Determine what PPTX metadata controls the output filename/path
• Catalog the image formats that python-pptx can extract (PNG, JPEG, EMF, WMF, TIFF, SVG, etc.)
• Assess maximum reasonable image sizes for presentation content

task-planner

• Define size limits and content validation strategy
• Plan path sanitization approach (os.path.realpath containment check)
• Determine if content-type verification is sufficient or if magic byte checking is needed

task-implementor

• Add blob size validation before write
• Add output path containment check (resolved path must be under expected directory)
• Add content-type / magic byte verification for known image formats
• Add tests with oversized blobs, path traversal filenames, and non-image content
• Sanitize any filename derived from PPTX metadata

Acceptance Criteria

• Blob size is validated against a configurable maximum before writing to disk
• Output paths are verified to be within the expected output directory (no path traversal)
• Content validation confirms blob data matches expected image formats
• Filenames derived from PPTX metadata are sanitized (no path separators, null bytes, or special characters)
• Tests verify rejection of oversized blobs, path traversal attempts, and non-image content
• No raw blob data is written to arbitrary locations without validation