analytics-architect

Pass

Audited by Gen Agent Trust Hub on Aug 27, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external data via the BrowserOS interface, which represents a potential attack surface for indirect instructions.
  • Ingestion points: External web content and data enter the agent context through BrowserOS.
  • Boundary markers: The instructions do not define technical delimiters for untrusted data, but they mandate that such content be treated as hostile.
  • Capability inventory: The skill possesses terminal execution capabilities via the ExecutionProxy (e.g., rtk npm test) and file-writing capabilities through the MemoryStore interface.
  • Sanitization: The protocol includes a mandatory requirement to redact secrets and PII from external data before sharing context with subagents, which acts as a mitigation against data exfiltration and context poisoning.
  • [SAFE]: The framework instructions emphasize least privilege and deterministic execution protocols. No evidence of malicious command execution, unauthorized network operations, or obfuscated content was found within the skill files.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 27, 2026, 07:35 AM
Security Audit — agent-trust-hub — analytics-architect