The Agent Skills Directory

[SAFE]: The skill is purely instructional and provides frameworks for B2B sales assets without executing code, scripts, or making network requests.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an ingestion surface for untrusted data via local context files such as .agents/product-marketing-context.md (referenced in SKILL.md). Boundary markers are present in the instructions, which explicitly state that local context files are background information and do not have the authority to override user instructions. A capability inventory across all files confirms the absence of subprocess calls, exec/eval functions, file-writing operations, or network requests. While sanitization is not explicitly defined, the complete lack of exploitable capabilities ensures that this ingestion surface is safe.

sales-enablement