ClawHub
Back to skill

Security audit

Doorstep

Security checks across malware telemetry and agentic risk

Overview

Doorstep is a disclosed San Francisco errand service integration, but it handles sensitive account, location, task, and payment-related details.

Install only if you are comfortable sharing errand details with Doorstep and human taskers. Use the minimum necessary address, timing, item, phone, and message details; be cautious with prescriptions or other sensitive requests; use a unique password or OAuth/browser signup where possible; protect the API key; and review the final task, price, address, timing, and any auto-approval limits before allowing payment or dispatch.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (2)

Missing User Warnings

Medium
Confidence
89% confidence
Finding
The README explicitly encourages tasks involving prescription pickup and real-time task tracking, which can require users to share health-related details, home/work addresses, and live location information with the service and human taskers. Presenting these examples without any privacy warning, data-handling disclosure, or guidance to minimize sensitive information increases the risk of unnecessary exposure of personal and potentially regulated data.

Missing User Warnings

Medium
Confidence
85% confidence
Finding
The skill instructs the agent to collect a user's email and password and submit them to an external registration service, but it does not warn that these are sensitive credentials being transmitted off-platform. This creates a real risk of credential exposure or unsafe handling by the agent or user, especially because the flow normalizes password collection inside the skill rather than redirecting the user to a trusted sign-up surface.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.