launch-ad-campaign
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local Python scripts (
approval-manager.pyandexecution-tracker.py) to manage approval records and log campaign lifecycle events. These commands use environment variables likeCLAUDE_PLUGIN_ROOTto locate scripts within the plugin environment. - [INDIRECT_PROMPT_INJECTION]: The skill processes data from external sources including brand profiles, compliance guidelines, and agency SOPs to configure campaign targeting and bid strategies.
- Ingestion points: Data is read from
~/.claude-marketing/brands/,~/.claude-marketing/sops/, and through instructions delegated from other campaign management skills. - Boundary markers: The skill implements a mandatory 'Execution Gate' that presents a complete summary of all proposed changes and requires the user to type 'yes' to proceed; any ambiguous or absent approval cancels the execution.
- Capability inventory: The skill has the ability to create live ad campaigns via connected MCP servers and run local scripts for tracking and approval management.
- Sanitization: The primary defense against malicious instructions embedded in profile data or SOPs is the mandatory human review step, where all generated campaign parameters are displayed for verification before deployment.
Audit Metadata