wix-headless-kit
Pass
Audited by Gen Agent Trust Hub on Nov 19, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill fetches site templates, code layers, and project scaffolds from the official Wix GitHub repository (github.com/wix/skills.git). These are verified vendor resources used to populate the local development environment.
- [COMMAND_EXECUTION]: The skill executes standard development and project management commands, including
npm install,npx @wix/cli, andgit clone. These commands are used to manage project dependencies and interact with the Wix platform's developer tools. - [INDIRECT_PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection by ingesting untrusted user-supplied data (such as CSV files, PDFs, or text lists) to generate site content plans (
plan.json). - Ingestion points: User-provided content files processed during the 'seed plan' step (SKILL.md).
- Boundary markers: Absent for raw data ingestion; the skill relies on the agent to correctly map unstructured or semi-structured data into a validated JSON schema.
- Capability inventory: The skill possesses capabilities to write files, execute subprocesses (npm/npx), and make authenticated API calls (
ExecuteWixAPI) to modify site content. - Sanitization: Not explicitly implemented in the skill scripts for raw user input; the security model relies on the agent's internal instruction-following and the platform's API-level validation.
- [DATA_EXFILTRATION]: The skill provides a mechanism to send user feedback and session telemetry to Wix API endpoints (
wixapis.com). This operation uses official Wix CLI tokens and is a documented vendor feature designed to improve the developer experience.
Audit Metadata