1c-project-init
Fail
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: HIGHCREDENTIALS_UNSAFECOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPERSISTENCEPRIVILEGE_ESCALATIONDATA_EXFILTRATION
Full Analysis
- [CREDENTIALS_UNSAFE]: The skill utilizes hardcoded credentials ('admin:admin123') when configuring Git remotes and directs the agent to store administrative passwords in plaintext within local configuration files like .mcp.json and .v8-project.json.\n- [COMMAND_EXECUTION]: The skill executes various local PowerShell scripts and remote shell commands via SSH to automate environment tasks and Docker container management.\n- [PRIVILEGE_ESCALATION]: Administrative operations are performed using 'ssh root@' access, which grants the agent full administrative control over the target remote server.\n- [PERSISTENCE]: The skill establishes a persistence mechanism by modifying the crontab on a remote server to schedule recurring Git synchronization tasks.\n- [REMOTE_CODE_EXECUTION]: The skill uses SSH to trigger remote executions, including Docker commands and custom shell scripts, on external servers.\n- [DATA_EXFILTRATION]: The initialization script reads sensitive API tokens from the user's local filesystem (~/.claude/private/credentials.md) and transmits them to a remote Gitea service for repository creation.
Recommendations
- AI detected serious security threats
Audit Metadata