The Agent Skills Directory

[DATA_EXFILTRATION]: The skill provides templates for managing sensitive configuration keys such as GITHUB_TOKEN and SUPABASE_SERVICE_ROLE_KEY. It identifies these as secrets and correctly recommends using platform-native secret management (e.g., Render's fromSecret) instead of hardcoding values.
[COMMAND_EXECUTION]: Documentation includes standard shell commands for configuration validation and linting (e.g., yamllint, hadolint, shellcheck, docker build). These are restricted to local environment syntax checks and artifact building.
[EXTERNAL_DOWNLOADS]: Dockerfile templates install standard system utilities (curl, git, jq) from official repositories and reference trusted base images (python:3.11-slim).
[PROMPT_INJECTION]: The skill is designed to parse configuration files. It specifically employs yaml.safe_load() for validation, which prevents the execution of arbitrary Python objects during YAML parsing.

configuration-file-parsing