The Agent Skills Directory

[SAFE]: The skill provides defensive configuration guidelines for Elixir/Phoenix deployments, aimed at preventing common configuration errors that lead to production outages.
[CREDENTIALS_SAFE]: The skill actively discourages hardcoding secrets in source code and provides instructions on using environment variables via Elixir's System.fetch_env!, which is a standard and recommended security practice.
[DATA_EXPOSURE]: The skill includes specific rules to set production log levels to :info. This is a defensive measure to prevent the leakage of Personally Identifiable Information (PII) and sensitive database query parameters that are typically logged at the :debug level in the Phoenix framework.
[INDIRECT_PROMPT_INJECTION]: The skill targets configuration files (**/config/*.exs) and deployment scripts. While it instructs the agent to analyze and potentially modify these files, it does so to enforce safety constraints and best practices. No exploitable surfaces or malicious instructions were found.

deployment-gotchas