skills/vercel-labs/dev3000/d3k/Gen Agent Trust Hub

d3k

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPRIVILEGE_ESCALATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • Global Package Installation: The skill installs the dev3000 package globally using bun or npm. This is a standard procedure for CLI-based development tools to ensure they are available in the user's environment.
  • Administrative Setup for Routing: The d3k portless setup command initiates an operating system-level authorization dialog to configure canonical HTTPS routing. This process requires explicit user approval through the system's native security prompts and is used to manage local DNS or host file configurations.
  • Indirect Prompt Injection Surface: The agent processes browser console logs and server output via commands like d3k logs and d3k errors. As these logs can contain content from external web pages, there is a potential surface for indirect prompt injection if an attacker-controlled site emits malicious instructions. The risk is minimized by the agent's focus on technical debugging evidence.
  • Ingestion points: Browser console logs and server logs retrieved via d3k logs and d3k errors --context (SKILL.md).
  • Boundary markers: None explicitly defined in the instructions for log processing.
  • Capability inventory: The skill can drive browser interactions, capture screenshots, and read/write files within the project directory and ~/.d3k/ (SKILL.md).
  • Sanitization: No specific sanitization or filtering of log content is described.
  • Managed Browser Interaction: The skill uses a project-specific Chrome profile to maintain session state (like authentication). It automates browser actions such as clicking and filling forms through the d3k agent-browser utility, which is confined to the managed runtime environment.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 10:10 AM
Security Audit — agent-trust-hub — d3k