The Agent Skills Directory

[SAFE]: The skill reads branding and voice guidelines from local files (social-context.md, .agents/social-context.md). This access is intended for establishing content context and does not target sensitive system paths, credentials, or environment variables.
[SAFE]: No executable code, shell commands, or external dependencies are present in the skill instructions or referenced files.
[SAFE]: The skill does not include any network-enabled tools or instructions that could facilitate data exfiltration.

linkedin-post