campaign-analytics
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external campaign data, including email copy from sending platforms and CRM exports. This creates a surface for indirect prompt injection where malicious instructions embedded in the processed data could influence the agent's behavior.
- Ingestion points: Data is imported from CSV, Google Sheets, or CRM exports (SKILL.md, Phase 2).
- Boundary markers: None identified in the prompt instructions to separate untrusted data from instructions.
- Capability inventory: The skill uses a local script (scripts/check-output.py) for verification and performs complex analysis and report generation.
- Sanitization: No explicit sanitization or filtering of campaign copy is mentioned before processing.
- [DATA_EXFILTRATION]: The skill requests access to highly sensitive business information, including total marketing spend, customer pipeline values, and CRM records. While this is aligned with the skill's primary purpose of ROI analysis, users should ensure the agent environment is secure when providing these data sources.
- Evidence: Intake questions in SKILL.md Phase 1 request budget context, CRM access, and detailed pipeline metrics.
Audit Metadata