psd-layer-reader
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted PSD files and extracts their layer names and text content into a JSON format intended for agent consumption.
- Ingestion points: The
scripts/psd_layers.pyscript opens and reads external Photoshop files provided by the user or an external environment. - Boundary markers: The output JSON contains raw string data from the PSD (e.g.,
originalName,textInfo.text) without delimiters or warnings for downstream consumers that the content may contain instructions. - Capability inventory: The skill writes parsed data to local files (specified by
--output). - Sanitization: While
scripts/utils.pyprovides name sanitization for structural use (name), theoriginalNameandtextInfofields preserve the raw content of the PSD file, which could contain malicious instructions. - [EXTERNAL_DOWNLOADS]: The skill requires the installation of a third-party Python library from the official package registry.
- Evidence:
SKILL.mdinstructs the user to runpy -3 -m pip install psd-toolsto handle PSD parsing.
Audit Metadata