kelly-ads
Pass
Audited by Gen Agent Trust Hub on Aug 25, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill instructions follow safety best practices and do not contain any patterns intended to bypass agent constraints or override core behavior.
- [DATA_EXFILTRATION]: The skill handles sensitive ad performance data exclusively within the Busabase environment. No hardcoded credentials or unauthorized data transmission to external domains were found. Environment variables are correctly used for platform authentication.
- [REMOTE_CODE_EXECUTION]: The skill does not perform any remote code execution. All scripts are local and dependencies are pinned to specific versions of well-known libraries like
busabase-sdkandhono. - [COMMAND_EXECUTION]: Command execution is limited to specific maintenance and data ingestion scripts (
ingest_reports.mjs,run_checks.mjs,execute_decisions.mjs) which are intended for use by the agent or operator to manage the dashboard state. These do not execute arbitrary user input. - [INDIRECT_PROMPT_INJECTION]: The skill processes external data from ad platform reports. While it ingests untrusted strings (e.g., campaign names, search terms), it employs HTML escaping in the UI and treats numeric data with appropriate parsing, minimizing the risk of indirect injection. The human-in-the-loop boundary ensures no mutations happen without review.
Audit Metadata