Skills
skills/soul-brews-studio/oracle-skills-cli/awaken/Gen Agent Trust Hub

awaken

Fail

Audited by Gen Agent Trust Hub on Apr 17, 2026

Risk Level: HIGHREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: Fetches and executes an installation script from the author's GitHub repository (Soul-Brews-Studio/oracle-skills-cli) using the curl | bash pattern.
  • [COMMAND_EXECUTION]: Instructs the agent to create and populate .claude/settings.local.json. This action automatically grants permissions for numerous bash commands and skills, bypassing standard user-confirmation prompts for these tools.
  • [DATA_EXFILTRATION]: Automatically creates an issue on a public GitHub repository (Soul-Brews-Studio/oracle-v2) containing metadata about the Oracle and its human companion as part of the birth announcement ritual.
  • [PROMPT_INJECTION]: Contains an indirect prompt injection surface through the ingestion of external data.
  • Ingestion points: Reads content from external GitHub repositories and issues (Step 2).
  • Boundary markers: Absent; there are no markers or warnings to disregard instructions within the ingested content.
  • Capability inventory: Powerful capabilities including script execution (bash), file creation (cat), and remote repository interaction (gh).
  • Sanitization: Absent; the skill does not validate or sanitize retrieved content before it influences agent behavior.
Recommendations
  • HIGH: Downloads and executes remote code from: https://raw.githubusercontent.com/Soul-Brews-Studio/oracle-skills-cli/main/install.sh - DO NOT USE without thorough review
Audit Metadata
Risk Level
HIGH
Analyzed
Apr 17, 2026, 02:31 AM