The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches configuration, release manifests, and plugin assets from GitHub (github.com and githubusercontent.com) and the vendor's own domains (bymiles.ai). These are established and expected sources for the skill's functionality.
[COMMAND_EXECUTION]: The skill executes a bundled shell launcher (scripts/miles) which in turn executes either a local runtime binary or a Node.js CLI script. It also supports using WP-CLI via an adapter to manage local WordPress installations.
[PRIVILEGE_ESCALATION]: The skill uses chmod +x during initialization to ensure its bundled CLI script is executable. This is standard practice for distributing scripts within a skill and does not target system-level files.
[DATA_EXFILTRATION]: The skill transmits site descriptions, design briefs, and brand assets to the bymiles.ai API. It also manages an API key stored locally in ~/.miles/credentials.json. These operations are the primary function of the skill for site generation and authentication.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect injection as it processes user-provided content (files, briefs) and relays it to an external API. However, the skill includes explicit instructions for the agent to relay this information verbatim without interpretation, which serves as a fidelity guardrail.

miles