security-reviewer

Pass

Audited by Gen Agent Trust Hub on Aug 27, 2026

Risk Level: SAFE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill explicitly acknowledges the risks of processing untrusted data. It mandates that web content and external data (ingested via 'BrowserOS') be treated as hostile. It provides a remediation strategy by requiring agents to redact secrets and PII before sharing context with subagents, demonstrating good security hygiene for handling indirect injection surfaces.
  • [REMOTE_CODE_EXECUTION]: No remote code execution patterns were found. The skill references an 'ExecutionProxy Interface' (using the rtk prefix) for terminal actions, which is a standard method for scoped command execution within the framework rather than an arbitrary remote execution vector.
  • [CREDENTIALS_UNSAFE]: No hardcoded credentials or secrets were found. The instructions explicitly direct the user/agent to redact secrets before sharing context, aligning with secret management best practices.
  • [DATA_EXFILTRATION]: No network exfiltration patterns or unauthorized data access commands were detected. The skill uses a 'MemoryStore' interface to save artifacts to local project directories (e.g., docs/departments/).
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 27, 2026, 07:36 AM
Security Audit — agent-trust-hub — security-reviewer