adversarial-reviewer

Pass

Audited by Gen Agent Trust Hub on Aug 27, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external plans, features, and launch documentation, creating an attack surface where malicious input could attempt to influence the agent's behavior.
  • Ingestion points: Processes external inputs described as "plan, feature, or launch" during the review cycle (SKILL.md).
  • Boundary markers: Absent. There are no defined delimiters or specific instructions for the agent to isolate untrusted input from its core instructions.
  • Capability inventory: Access to project architecture files (docs/graph.json), library metadata (package.json), persistent documentation storage (docs/departments/), and terminal execution capabilities via the ExecutionProxy Interface (SKILL.md).
  • Sanitization: The instructions include proactive mitigation by mandating that external data be treated as hostile and that secrets or PII be redacted before context sharing (SKILL.md).
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 27, 2026, 07:35 AM
Security Audit — agent-trust-hub — adversarial-reviewer