hermes-agentmesh-async-bus

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). The workflow instructs the agent to configure and interact with a Redis message bus where remote workers process tasks and message content, which is a monitored message queue/feed.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.70). The skill instructs cloning an arbitrary third-party GitHub repository (seleman66eeddwegger3-art/hermes-agentmesh.git) and pulling installation scripts from an unverified local HTTP server address, constituting a weak/unknown operational dependency under the dependency gate.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 1.00). The skill prompt instructs the agent to clone an external repository and run installation steps that modify system configuration files using sudo (such as modifying Redis configuration in /opt/homebrew/etc/redis.conf), which compromises the machine's state and security controls.