marketing-os-starter
Warn
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill's setup instructions include cloning a project structure from
https://github.com/ericosiu/marketing-os-starter.git, which is an unverified source. - [COMMAND_EXECUTION]: The skill guides the agent and user through executing shell commands, including
git cloneand the global installation of NPM packages like@mcp/google-analyticsand@mcp/hubspotto enable data integration. - [INDIRECT_PROMPT_INJECTION]: The skill's architecture is built to ingest and process external content such as customer pain points and quotes from public sources.
- Ingestion points: Data is collected by a specialized agent and stored locally in the
memory/directory. - Boundary markers: The workflow utilizes structured JSON files for communication between agents, such as
audience-profile.jsonandcreative-brief.json. - Capability inventory: The agent possesses capabilities for file system manipulation and terminal command execution.
- Sanitization: The instructions do not specify any validation or sanitization protocols for the external data before it is utilized by other agents in the pipeline.
Audit Metadata