email-sequences

Pass

Audited by Gen Agent Trust Hub on Nov 17, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted data from Voice of Customer (voc.md) files and external Email Service Provider (ESP) API responses. This represents a potential surface for indirect prompt injection where malicious instructions could be embedded in data processed by the agent.
  • Ingestion points: voc.md, learnings.md, and API responses from Kit, Beehiiv, and Resend.
  • Boundary markers: The skill instructions do not specify strict delimiters or specialized prompt headers to isolate untrusted data.
  • Capability inventory: The skill has file-system write access and the ability to execute shell scripts for domain auditing.
  • Sanitization: There is no evidence of specialized sanitization for external content before it is processed for sequence generation.
  • [COMMAND_EXECUTION]: The skill utilizes local shell and Python scripts (deliverability_check.sh and brand_context.py) to perform project configuration and DNS deliverability checks. The scripts are contained within the skill's own directory structure and use secure coding practices such as variable quoting to handle user input safely.
  • [EXTERNAL_DOWNLOADS]: The skill connects to established technology platforms including Kit, Beehiiv, and Resend to manage email campaigns. These connections are used for legitimate data transfer (fetching stats and pushing drafts) and are gated by explicit user confirmation requirements.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 17, 2026, 04:01 PM