The Agent Skills Directory

[SAFE]: The skill utilizes a sophisticated state-management system localized to the .hex-skills/ directory. This system tracks progress through various worker phases using structured JSON state and history files, ensuring project-scoped execution without external dependency for state persistence.
[SAFE]: Detailed security practices for destructive operations are documented in references/destructive_operation_safety.md. This reference mandates backup plans, rollback procedures, blast radius assessment, and environment guards for sensitive operations like database drops or file system deletions.
[SAFE]: The runtime scripts included with the skill (under references/scripts/) follow secure coding practices, utilizing standard Node.js built-in modules (node:fs, node:path, node:crypto) and implementing atomic writes and JSON schema validation for all state updates.
[SAFE]: External resource management for templates (described in references/template_loading_pattern.md) follows a project-independence model where templates are copied to the target project upon first use, allowing for local customization and preventing dependency on remote sources at runtime.
[SAFE]: No hardcoded credentials, suspicious network exfiltration, or obfuscated patterns were identified. The skill's architecture emphasizes transparency and adherence to strict validation rules (e.g., creation_quality_checklist.md).

ln-301-task-creator