skills/joelhooks/joelclaw/talon/Gen Agent Trust Hub

talon

Pass

Audited by Gen Agent Trust Hub on Mar 5, 2026

Risk Level: SAFECOMMAND_EXECUTIONREMOTE_CODE_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill facilitates the execution of powerful administrative commands including launchctl for service management, kubectl for Kubernetes orchestration, and docker for container operations. These are necessary for the tool's primary purpose of infrastructure supervision.
  • [REMOTE_CODE_EXECUTION]: Several troubleshooting examples use pipes to format output, such as curl -sS http://127.0.0.1:9999/health | python3 -m json.tool. While piping network data to an interpreter is a common security concern, in this context it is used for local diagnostic formatting via a standard library module.
  • [COMMAND_EXECUTION]: The watchdog performs health probes on remote virtual machines (Colima) by executing commands over SSH (ssh -F ...). This is used to detect 'split-brain' scenarios between the host and VM environment.
  • [DATA_EXFILTRATION]: The escalation system is designed to send 'SOS' notifications to external services like Telegram and iMessage. This represents an intentional data flow for incident alerting and utilizes a secure secrets-leasing mechanism for credentials.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 5, 2026, 10:20 PM