anycap-social-meme-workflows

Pass

Audited by Gen Agent Trust Hub on Apr 24, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill relies on a local CLI tool named 'anycap' to perform image generation and analysis tasks. It executes shell commands like 'anycap image generate' and 'anycap actions image-read' to automate the meme workflow. These commands are integral to the skill's purpose and are documented as expected vendor tools.
  • [PROMPT_INJECTION]: Potential for indirect prompt injection exists during image processing steps where untrusted data is ingested.
  • Ingestion points: Untrusted content enters the workflow via 'anycap image generate' (using the '--param images=./source.png' flag) and through the 'anycap actions image-read' command which inspects './final-meme.png' for QA purposes.
  • Boundary markers: Absent; there are no specific markers or instructions to the model to ignore embedded visual or metadata-based instructions within the images being processed.
  • Capability inventory: The skill possesses the capability to execute shell commands via the 'anycap' CLI to create files, modify images, and query model schemas across various scripts.
  • Sanitization: Absent; no sanitization, validation, or filtering is described for external image content before it is passed to the analysis or generation tools.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 24, 2026, 08:44 AM