The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands via the Wix CLI (npx @wix/cli@latest) and a local Node.js script (scripts/bootstrap.mjs) to handle authentication, site initialization, building, and deployment. These operations are consistent with the skill's stated purpose of managing Wix projects.
[EXTERNAL_DOWNLOADS]: The skill uses npx @wix/cli@latest, which downloads the latest version of the Wix CLI from the npm registry. It also fetches configuration and documentation from official Wix domains (dev.wix.com, wixapis.com). These are well-known and trusted sources.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it reads data from Wix APIs (e.g., product descriptions, blog posts, CMS content) and provides instructions to the agent based on this data. The risk is managed by the skill's use of specific implementation recipes and structured data handling.
Ingestion points: references/DISCOVERY.md (reads site context via dynamic-context API), references/inline-recipes/ (reads existing content like campaigns, products).
Boundary markers: Not explicitly defined in the prompts to delimit untrusted API data.
Capability inventory: Shell execution via scripts/bootstrap.mjs and the Wix CLI, network operations via curl to Wix APIs.
Sanitization: Not explicitly implemented for external content before interpolation.

wix-headless