The Agent Skills Directory

[SAFE]: The skill follows a modular architecture where an orchestrator coordinates specialized sub-agents (Format, VoC, Hook, Body, CTA, Compliance, Variant, Critic) to produce marketing content. This structure promotes clear separation of concerns and quality control.
[SAFE]: Data processing involves reading marketing research artifacts (like IMC plans and ICP research) from a local .agents/ directory. While these inputs enter the agent context, the sub-agent instructions are highly task-specific and include specific output contracts, which mitigates the risk of indirect prompt injection.
[SAFE]: Tool usage is restricted and aligns with the skill's stated purpose. Tools like WebSearch and WebFetch are employed by the Format Agent for verifying platform specifications, and no dangerous system commands or persistence mechanisms were found.
[SAFE]: There are no hardcoded credentials, sensitive file access patterns (e.g., targeting .ssh or .aws directories), or obfuscated payloads in any of the provided skill files.

content-create