inbox-commander

Pass

Audited by Gen Agent Trust Hub on Feb 19, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • [Indirect Prompt Injection] (LOW): This skill has an attack surface for indirect prompt injection because it processes untrusted data (external emails) using external reference files (references/known-senders.md).
  • Ingestion points: The skill ingests untrusted email content from external senders.
  • Boundary markers: None identified in the provided reference files.
  • Capability inventory: No executable capabilities (subprocess, eval, file-write, network) are present in the provided files.
  • Sanitization: None identified; the system relies on pattern matching defined in markdown files.
  • [Metadata Poisoning] (SAFE): The files contain configuration templates and examples (e.g., VIP contacts, vendors) which are standard for an email triage skill and do not contain deceptive instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 19, 2026, 01:20 AM