The Agent Skills Directory

[SAFE]: The skill uses tools like Bash, Read, and Write for legitimate path resolution and project artifact management. All file operations are constrained to the local environment and project-specific directories defined in configuration files like projects.yaml and .claude/paths.md. No network activity or credential harvesting was identified.
[SAFE]: The skill mentions external design inspirations (e.g., from entpnomad/tone-of-voice), but these are attributions and do not involve runtime downloads or execution of external code.
[SAFE]: Indirect prompt injection risk is minimal as the skill performs linguistic analysis on provided samples rather than executing embedded instructions within those samples. The presence of a 'Garde-fous' (guardrails) section provides additional structural control over generated output.

tune-voice