The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references installing ngrok via Homebrew (brew install ngrok) to enable public access for media imports. This is a well-known technology service used according to its standard operational purpose.
[COMMAND_EXECUTION]: The skill executes locally defined scripts from sibling adapter skills (e.g., rp-source-wordpress/scripts/) and runs ngrok commands for network tunneling. These commands are integral to the migration workflow.
[PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it processes external, potentially untrusted data like source API documentation and data exports. This risk is inherent to the migration use case and is documented as part of the operational logic.
Ingestion points: Processes source API docs, exports, and local dump files.
Boundary markers: No explicit delimiters for untrusted content are defined.
Capability inventory: Performs file writes and executes subprocesses for adapter scripts and ngrok.
Sanitization: No specific data validation or sanitization of the source inputs is performed.
[DATA_EXFILTRATION]: The skill includes explicit security instructions for handling .env files, directing the agent to verify the existence of keys like WIX_AUTH_TOKEN without reading or echoing the secret values, which prevents accidental credential exposure.

rp-discovery