account-lifecycle-summaries
Pass
Audited by Gen Agent Trust Hub on Mar 3, 2026
Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
- [PROMPT_INJECTION]: The skill is subject to indirect prompt injection risks as it is designed to ingest and process untrusted external data.\n
- Ingestion points: Ingests transaction history, service interactions, and account data from core system extracts (documented in SKILL.md).\n
- Boundary markers: The methodology lacks explicit delimiters or specific instructions for the agent to disregard instructions potentially embedded within the source data records.\n
- Capability inventory: The skill is limited to narrative generation and does not have access to subprocess execution, network operations, or file system writes.\n
- Sanitization: While PII masking for account numbers is recommended, no sanitization is present to detect or neutralize malicious commands hidden in transaction ledgers or CRM logs.\n- [NO_CODE]: No executable scripts or binary files are included with the skill; it relies entirely on markdown-based instructions for the AI agent.
Audit Metadata