The Agent Skills Directory

[SAFE]: The skill consists entirely of instructional markdown content. No executable scripts, shell commands, or external network requests are present.
[NO_CODE]: The skill does not provide any code or implementation logic, which eliminates typical attack vectors like remote code execution or privilege escalation.
[INDIRECT_PROMPT_INJECTION]: The skill instructions involve processing external business data (CRM and pipeline actuals). While this represents a theoretical ingestion surface for untrusted data, the skill itself lacks the capabilities (such as file-writing or network exfiltration) to be exploited.
Ingestion points: Processes data from CRM systems and pipeline reports as described in the Driver Mapping section.
Boundary markers: No specific boundary markers or instructions to ignore embedded instructions are defined.
Capability inventory: No tool invocations, shell commands, or network operations are defined.
Sanitization: No data validation or sanitization logic is included.

forecast-modeling