web3-smart-contracts
Pass
Audited by Gen Agent Trust Hub on Apr 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill instructs the agent to modify its default behavior by beginning its first response with a specific emoji (🧢).
- [COMMAND_EXECUTION]: The 'Companion check' section directs the agent to execute a shell command (
ls) to inspect specific hidden directories in the user's home folder (e.g.,~/.claude/skills/) to identify which companion skills from the same author are currently installed. - [DATA_EXPOSURE & EXFILTRATION]: While the skill accesses hidden directory structures, the operation is limited to listing files in known configuration paths for the purpose of dependency management and does not target sensitive user data like credentials or SSH keys.
Audit Metadata