The Agent Skills Directory

[COMMAND_EXECUTION]: The skill mentions common Phoenix CLI tools like mix phx.new, mix phx.gen.live, and mix phx.gen.auth. These are standard development commands for generating application scaffolding and do not involve arbitrary or dangerous command execution.
[CREDENTIALS_UNSAFE]: The skill explicitly instructs developers to use config/runtime.exs and environment variables for secrets, specifically warning against hardcoding credentials. This follows industry security standards.
[DATA_EXFILTRATION]: No network operations or unauthorized data access patterns were found. The skill reinforces data ownership by requiring a Scope (user/org/tenant) for all context functions to prevent cross-tenant data leakage.
[PROMPT_INJECTION]: The instructions are purely technical and do not contain language intended to override agent safety filters or bypass system instructions.
[REMOTE_CODE_EXECUTION]: There are no patterns involving the download or execution of remote scripts or unverified dependencies.
[PRIVILEGE_ESCALATION]: The skill focuses on application-level authorization (Authz) and does not request or attempt to acquire elevated system privileges (e.g., sudo, chmod 777).
[OBFUSCATION]: The content is written in clear, standard technical Portuguese and Elixir code. No hidden characters, encoded URLs, or homoglyph attacks were detected.
[INDIRECT_PROMPT_INJECTION]: While the skill describes how to build web applications that ingest user data, it reinforces security boundaries by recommending the use of Changesets for input validation and Scopes to ensure data isolation at the domain layer.

phoenix