strip-image-ai-metadata
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data (images and metadata files) which could potentially contain malicious instructions targeting the agent or the underlying parsers.\n
- Ingestion points: The script
scripts/strip_c2pa.pyreads and parses binary image data and structured XMP (XML) and JSON sidecar files.\n - Boundary markers: There are no explicit delimiters or specific instructions for the agent to disregard natural language content found within the metadata of the processed files.\n
- Capability inventory: The skill possesses the capability to write and overwrite files on the local filesystem using atomic replacement logic.\n
- Sanitization: The script performs significant sanitization by identify and removing AI-related metadata chunks, EXIF tags, and XMP properties based on a comprehensive list of keywords and specific binary signatures, effectively reducing the risk of malicious instructions persisting in the output.
Audit Metadata