hostc-public-preview

Pass

Audited by Gen Agent Trust Hub on May 11, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill uses npx hostc@latest to download and execute the hostc utility from the npm registry at runtime.
  • [COMMAND_EXECUTION]: The instructions direct the agent to execute shell commands for starting and configuring the public tunnel.
  • [PROMPT_INJECTION]: The skill requires the agent to monitor development server output to infer port numbers, which presents a surface for indirect prompt injection if the output is controlled by an untrusted source.
  • Ingestion points: Development server console output.
  • Boundary markers: Absent.
  • Capability inventory: Command execution and network tunneling via npx hostc.
  • Sanitization: Absent.
Audit Metadata
Risk Level
SAFE
Analyzed
May 11, 2026, 03:06 PM
Security Audit — agent-trust-hub — hostc-public-preview