amir-publish

Warn

Audited by Gen Agent Trust Hub on Jun 16, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONREMOTE_CODE_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to perform extensive shell orchestration, including repository state inspection via git status, local staging via git add -u, and remote access using ssh.
  • [REMOTE_CODE_EXECUTION]: The skill triggers the execution of make install locally and on five hardcoded remote machines (amirs-m3-max-new, agents@amirs-mac-studio, amir-m5, home, and claw). Running make targets executes arbitrary instructions defined in the repository's Makefile, which are not visible within the skill itself.
  • [REMOTE_CODE_EXECUTION]: The skill facilitates remote execution by directing the agent to SSH into multiple hosts to fetch, checkout, and pull repository changes followed by an automated installation step. This chain allows for the propagation and execution of code across a network cluster.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jun 16, 2026, 09:09 AM
Security Audit — agent-trust-hub — amir-publish