muapi-chibi-collage-effect
Pass
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to use the
muapiCLI tool to perform image editing tasks. It also provides a fallback tocurlfor direct API interaction if CLI aliases are missing, which involves executing shell commands to transmit data and API keys. - [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied image URLs (
person_image) to drive its core logic. The agent is instructed to analyze the image content to determine themes, poses, and captions. - Ingestion points: The
person_imageinput field inSKILL.mdallows users to provide external URLs. - Boundary markers: None are present to delimit the analysis of the external image content from the agent's core instructions.
- Capability inventory: The skill utilizes
muapiCLI execution andcurlfor network operations to send data to external endpoints. - Sanitization: There is no evidence of sanitization or validation of the content retrieved from the provided image URLs before processing.
- [DATA_EXFILTRATION]: The skill performs network operations to
api.muapi.aito process images and requires the transmission of aMUAPI_API_KEY. While this is the intended functionality, it involves sending user-provided data to a third-party service.
Audit Metadata