kelly-brand
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill includes several utility scripts (
setup.mjs,publish_airapp.mjs,execute_decisions.mjs) designed for infrastructure management. These scripts are used to provision Busabase bases, deploy the companion UI code, and promote approved narrative items. They require explicit environment variables for authentication and are standard for the skill's architecture. - [EXTERNAL_DOWNLOADS]: The skill uses the
pnpmpackage manager to manage dependencies, including established libraries such asbusabase-sdkandhono. No unverified or suspicious remote sources were detected. - [INDIRECT_PROMPT_INJECTION]: The drift monitoring feature involves ingesting content from external channels (e.g., websites, social media) to check for brand alignment. This introduces a surface for indirect prompt injection. The skill mitigates this by requiring mandatory human review and implementing an automated quality audit gate (
narrative-quality-auditor) that scores and labels content before it can be adopted into the canonical narrative. - [DATA_EXPOSURE]: The skill explicitly instructs the agent to avoid asking for secrets in chat and uses standard environment variable practices for its administrative scripts. Data is stored in the user's Busabase workspace.
Audit Metadata