muapi-brand-kit
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONDATA_EXFILTRATIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied branding data without adequate isolation or sanitization.\n
- Ingestion points: brand_name, industry, personality, and color_preference inputs defined in SKILL.md.\n
- Boundary markers: Absent; inputs are interpolated directly into prompts like "Minimalist logo for '{{brand_name}}'...".\n
- Capability inventory: Shell command execution (muapi CLI) and network requests (curl).\n
- Sanitization: No instructions are provided for escaping or validating input variables before use.\n- [COMMAND_EXECUTION]: The executing agent is instructed to construct and run shell commands (muapi and curl) by substituting placeholders with raw user input, which could lead to command injection if inputs are not properly handled by the agent.\n- [DATA_EXFILTRATION]: The skill uses curl to communicate with the external domain api.muapi.ai to transmit brand details and retrieve generated assets.\n- [DYNAMIC_EXECUTION]: CLI commands and API payloads are dynamically assembled at runtime based on user-provided parameters.
Audit Metadata