figma-skill
Fail
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: HIGHREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill is a stub that uses
npx skills addto download and install a full code payload from an unverified GitHub repository (nafiurrahmanniloy/figma-skill). The skill documentation explicitly mentions that it includes executable Python scripts, which will be run on the user's system once the external payload is installed. - [EXTERNAL_DOWNLOADS]: The skill triggers the download of instructions and executable scripts from an external GitHub account. Furthermore, its core functionality involves downloading image assets and design tokens from external Figma sources.
- [COMMAND_EXECUTION]: The
SKILL.mdfile contains several shell commands designed to be executed by the agent to detect file paths, check installation status, and perform the installation of the remote skill package vianpx. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted data from external sources, specifically Figma Community resources and project files. There are no boundary markers or sanitization instructions provided in the stub to prevent malicious instructions embedded in these third-party designs from influencing the agent's behavior during code generation or token extraction.
Recommendations
- AI detected serious security threats
Audit Metadata