The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is instructed to run node on generated JavaScript artifacts located in the migrations/<project>/src/ directory. Executing code generated or stored within the project workspace is a dynamic execution pattern that requires verifying the integrity of the source scripts.
[CREDENTIALS_UNSAFE]: The workflow involves managing sensitive credentials, including WIX_AUTH_TOKEN and WP_APPLICATION_PASSWORD, which are stored in .env files. Although the skill includes explicit instructions to mask these secrets and avoid whole-file reads, the reliance on plain-text configuration for API tokens is a significant security consideration.
[EXTERNAL_DOWNLOADS]: To support media migration from local environments, the skill recommends the installation of ngrok via brew, which introduces a dependency on an external tunneling service.
[PROMPT_INJECTION]: The skill ingests untrusted project data from files like import-plan.md and setup-verification.md to inform its migration logic. This creates a surface for indirect prompt injection if the project files contain malicious instructions.
Ingestion points: Processes content from migrations/<project>/import-plan.md and setup-verification.md.
Boundary markers: No specific delimiters or safety instructions are defined to separate ingested data from agent commands.
Capability inventory: Includes shell command execution (node), file system modifications, and network access to www.wixapis.com.
Sanitization: No explicit sanitization or validation of the content within the ingested project files is mentioned.

rp-execute-import