algo-ad-budget
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes user-provided campaign metadata and historical performance data (spend, conversions, revenue) to fit response curves. This creates an attack surface where malicious instructions could be embedded in campaign identifiers or external data sources.
- Ingestion points: Historical campaign data points and campaign names collected in SKILL.md (Phase 1).
- Boundary markers: The instructions do not specify the use of delimiters or warnings to ignore instructions embedded in the historical data.
- Capability inventory: The skill defines mathematical optimization and curve fitting logic. It does not instruct or perform network exfiltration, file system modification, or subprocess execution.
- Sanitization: No input validation or sanitization rules are provided for the campaign identifiers.
Audit Metadata