meta-roi-framework
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection as it is designed to ingest and process data from external, untrusted sources which could contain malicious instructions.\n
- Ingestion points: Data entering the agent context includes client CRM records, financial invoices, inquiry logs, and GA4 exports, as specified in
SKILL.md(Required Inputs) andreferences/roi-model-method.md.\n - Boundary markers: The instructions lack explicit delimitation or directives to the agent to disregard instructions potentially embedded within the ingested data.\n
- Capability inventory: The skill utilizes search and read capabilities and anticipates downstream execution of broadcasts or customer outreach (referenced in
references/investment-business-case.md, Step 4).\n - Sanitization: No mechanisms for filtering, escaping, or validating the external content are described before the data is interpolated into the agent's workflow.
Audit Metadata