cold-email-onboarding

Pass

Audited by Gen Agent Trust Hub on Sep 8, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill serves as a data ingestion point for the cold-email framework, capturing a wide range of user-defined strings and saving them to brand-config.json and SOUL.md.
  • Ingestion points: Steps 1 through 7 involve collecting operator name, title, company details, target audience descriptions, and voice fingerprints directly from chat input.
  • Boundary markers: There are no explicit delimiters or boundary markers defined to isolate user-provided text within the resulting configuration files.
  • Capability inventory: The skill utilizes the Write tool to generate and update configuration files at the project root based on these inputs.
  • Sanitization: The skill does not provide evidence of sanitization or validation for the collected strings, creating a surface where malicious instructions could be stored and subsequently processed by other skills in the pack.
  • [EXTERNAL_DOWNLOADS]: The skill references and suggests the manual installation of companion skills from the vendor, such as cmj-hub/claude-psp and cmj-hub/claude-evp.
  • These references are presented as recommendations to the user to facilitate the completion of the onboarding process when specific data components are missing.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 8, 2026, 10:28 PM
Security Audit — agent-trust-hub — cold-email-onboarding