team-communication-protocols

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFE
Full Analysis
  • [Indirect Prompt Injection] (LOW): The skill establishes patterns for receiving and processing messages from other agents. While it uses structured JSON which helps with parsing, the potential for teammate-provided content to influence agent behavior represents a standard indirect injection surface common to multi-agent systems.
  • [Data Exposure] (SAFE): Teammate discovery relies on a specific local path (~/.claude/teams/) to identify agent metadata. This is limited to internal application configuration and does not involve searching for system secrets or user credentials.
  • [Command Execution] (SAFE): The skill is entirely instructional and does not include shell scripts, binary files, or commands that interact with the host operating system.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 04:38 PM