charles-arcade-assets
Pass
Audited by Gen Agent Trust Hub on Nov 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill makes extensive use of the
subprocessmodule to orchestrate media workflows. Specifically,onsets.py,words.py, andqa.pyinvokeffmpegfor audio analysis and video synchronization;source.pyusesnpx playwrightto automate browser screenshots; andnew_project.pycallsffprobefor media metadata. These executions are confined to internal utility calls necessary for the skill's primary function of asset generation and rendering. - [EXTERNAL_DOWNLOADS]: The
source.pyandfal.pyscripts fetch assets and interact with external APIs. Resources are sourced from well-known and reputable domains includingitunes.apple.com(Apple App Store info),wikimedia.org(creative commons photos),openlibrary.org(book metadata),simpleicons.org(brand logos), andfal.ai(generative AI models). All network operations are aligned with the skill's stated purpose of incorporating real-world references into graphics. - [INDIRECT_PROMPT_INJECTION]: A vulnerability surface is present because the skill extracts content from arbitrary user-provided URLs using browser automation in
source.py. This could theoretically be exploited by a malicious website to manipulate the data extracted by the agent. However, since the output is primarily graphical (screenshots and extracted images) and intended for manual review within a creative workflow, the immediate risk is low.
Audit Metadata