agentkits-marketing-automation
Warn
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the user to execute code from an unverified npm package (
@aitytech/agentkits-marketing) and clone a repository from an untrusted GitHub account. - [COMMAND_EXECUTION]: The installation process involves shell commands that modify the agent's internal configuration by copying files into
~/.claude/agents/,~/.claude/skills/, and other system directories. - [INDIRECT_PROMPT_INJECTION]: The skill features several commands that process untrusted data from external sources, creating an attack surface for indirect prompt injection.\n- Ingestion points: The
/seo:audit,/competitor:deep, and/cro:pagecommands ingest content directly from external URLs or user-provided HTML markup.\n- Boundary markers: The skill does not implement delimiters or provide instructions to the model to ignore instructions found within the processed data.\n- Capability inventory: The skill has file system write capabilities during installation and includes Python code snippets for interacting with CRM, analytics, and email service providers.\n- Sanitization: There is no evidence of sanitization or safety filtering for the data fetched from external web sources.
Audit Metadata