ProofTrail MCP Skill

• Outcome: Pass
• Latest audit: May 29, 2026 at 8:36 PM UTC
• Version:
  0.1.1

Overview

The skill appears to use normal, user-directed local setup steps for an MCP integration, with the main caveat that those steps run local code from the chosen repository.

Install only if you trust the repository source. Before running pnpm install or pnpm mcp:start, review the repo, package.json scripts, and lockfile when practical, and prefer a sandbox or non-sensitive machine for first use.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (1)

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The skill instructs the agent to clone a repo, run `pnpm install`, and start a local MCP server, which can execute untrusted code and install dependencies on the host. Because the skill does not explicitly warn about local code execution, dependency script execution, or the trust boundary of running a repo-local stdio server, an operator could follow these steps without understanding the security implications.

VirusTotal

66/66 vendors flagged this skill as clean.