prod-recon

Fail

Audited by Gen Agent Trust Hub on Nov 5, 2026

Risk Level: HIGHDATA_EXFILTRATIONPRIVILEGE_ESCALATIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [DATA_EXFILTRATION]: The skill directs the agent to access and read the ~/.ssh/config file to identify server aliases, hostnames, and identity files. It also instructs the agent to locate environment variable files (e.g., .env) within the project repository to discover deployment configuration.
  • [PRIVILEGE_ESCALATION]: The instructions explicitly permit and require the use of sudo apt-get install -y sshpass to install missing dependencies on the host system, which involves running commands with root privileges.
  • [COMMAND_EXECUTION]: The skill makes extensive use of shell commands for system discovery and interaction, including ssh, docker, sshpass, and standard Unix utilities like cat, grep, and sed. It specifically utilizes sshpass to handle passwords via environment variables passed to the sub-process.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from application logs and database query results (SKILL.md). It lacks boundary markers to delimit external content from instructions. The skill possesses significant capabilities, including executing shell commands (ssh, docker) and performing file system operations across the repository. No explicit sanitization or escaping of the ingested log or database content is performed prior to processing.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Nov 5, 2026, 10:40 PM