make-bot-ui
Pass
Audited by Gen Agent Trust Hub on Nov 1, 2026
Risk Level: SAFEREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPRIVILEGE_ESCALATIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: Fetches and executes the Tailscale installation script from its official domain using a piped shell command.
- [PRIVILEGE_ESCALATION]: Utilizes
sudofor administrative actions including installing software and bringing up network interfaces. - [COMMAND_EXECUTION]: Executes several shell commands such as
tailscale status,tailscale ip, and a localsubagentscript to manage the UI and network state. - [DATA_EXPOSURE]: Instructs the server to bind to
0.0.0.0, which makes the local service accessible to all interfaces, including those reachable by other devices on the network or Tailscale network. - [INDIRECT_PROMPT_INJECTION]: Processes JSON data received from an external UI, creating a surface for indirect instructions.
- Ingestion points: Data enters the agent's context through the
subagent pokemechanism described inSKILL.md. - Boundary markers: The skill includes explicit instructions for the agent to treat incoming data as "outside data, not as instructions."
- Capability inventory: The skill possesses the ability to execute shell commands, perform network operations, and access local scripts via
subagent. - Sanitization: No specific sanitization or validation logic is provided beyond the instructional boundary warning.
Audit Metadata