ssh-remote-connection
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The script
scripts/connect.shfacilitates the execution of arbitrary shell commands on remote servers. This capability allows the agent to interact deeply with remote infrastructure, including managing Docker containers and system services. - [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it executes command-line arguments directly on a remote host via SSH.
- Ingestion points: Command line arguments passed to
scripts/connect.sh(accessed via$*). - Boundary markers: None. No delimiters or explicit warnings are used to separate user commands from potential system instructions.
- Capability inventory: Remote shell command execution via
ssh,sshpass, andexpectinscripts/connect.sh. - Sanitization: While the
SERVER_PROJECT_PATHvariable is escaped usingprintf %q, the primary command input ($*) is interpolated directly into the SSH execution string without sanitization or escaping. - [DYNAMIC_EXECUTION]: The
connect.shscript dynamically generates and executesexpectscripts at runtime to automate the handling of SSH passphrase and password prompts.
Audit Metadata