roi-calculator
Pass
Audited by Gen Agent Trust Hub on Jul 7, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill follows best practices for a data-processing utility. It reads campaign inputs provided by the user and writes calculations to localized memory paths within the project structure (e.g.,
memory/influencer/roi-calculator/). It does not contain any hardcoded credentials, remote downloads, or hidden scripts. - [PROMPT_INJECTION]: The skill ingests untrusted data from the user (campaign spend, results, and metrics). This creates a surface for indirect prompt injection (Category 8); however, the instructions mandate the use of strictly defined templates from
references/roi-templates.md. By forcing the agent to populate specific markdown tables and formulas, the risk of external data being misinterpreted as system instructions is minimized.
Audit Metadata