The Agent Skills Directory

[SAFE]: The skill consists entirely of instructional markdown and evaluation data. No scripts, shell commands, or automated tools are included, posing no direct execution risk.
[INDIRECT_PROMPT_INJECTION]: The skill references loading a local context file (.agents/product-marketing-context.md). While this represents a data ingestion surface, the skill lacks any capabilities for command execution, file writing, or network communication, meaning it cannot be leveraged for data exfiltration or system modification.
[METADATA_POISONING]: The metadata fields (name, description, version) are descriptive and consistent with the skill's stated purpose, with no evidence of deceptive instructions or malicious intent.

site-architecture