day1-onboarding
Fail
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: HIGHREMOTE_CODE_EXECUTIONPERSISTENCECOMMAND_EXECUTIONDYNAMIC_EXECUTIONPRIVILEGE_ESCALATIONDYNAMIC_CONTEXT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill directs the agent to install plugins from the third-party 'obra/superpowers-marketplace', which facilitates the execution of unverified remote code.
- [PERSISTENCE]: The skill modifies agent configuration files ('~/.claude/settings.json' and '.claude/settings.local.json') to register event-based hooks and status line commands that execute scripts, ensuring certain operations persist across sessions.
- [COMMAND_EXECUTION]: The skill instructs the agent to run various shell commands and git operations to manage the local environment and demonstrates direct terminal access.
- [DYNAMIC_EXECUTION]: The skill provides a bash script for a custom status line and configures the agent to execute it automatically using the 'command' status line type.
- [PRIVILEGE_ESCALATION]: The agent is instructed to modify its own internal settings to enable experimental features, potentially bypassing standard security constraints or operational limits.
- [DYNAMIC_CONTEXT_INJECTION]: The skill demonstrates and contains usage of the '!command' syntax, which allows for shell command execution within the agent's processing context.
Recommendations
- AI detected serious security threats
Audit Metadata