advocacy-loops
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill contains an attack surface for indirect prompt injection via the ingestion of shared configuration data.
- Ingestion points: The instructions in
SKILL.mddirect the agent to read an external shared context file (.agents/marketing-context.md) and treat it as a source of truth and constraints. - Boundary markers: There are no defined delimiters or instructions to ignore potential embedded commands within the
marketing-context.mdfile. - Capability inventory: Analysis of the skill and its references (
references/referral-system.md) shows no use of subprocess execution, network operations, or file system write capabilities. The skill is limited to text generation and strategic design tasks. - Sanitization: The skill does not implement sanitization or filtering for the external context file content.
Audit Metadata