ClawHub

SPF Exec Plan

Security checks across malware telemetry and agentic risk

Overview

This plan-execution skill is not malicious, but it needs review because it can drive code changes while giving inconsistent workspace-safety guidance and requiring local workflow-state edits.

Install only if you want an agent to execute implementation plans with local file changes. Use a clean branch or worktree, review the plan and any referenced superpowers skills first, and require explicit approval before sub-agent delegation or updates to progress, findings, and handoff files.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

Intent-Code Divergence

Medium
Confidence
95% confidence
Finding
The skill contains directly conflicting isolation guidance: one section says using an isolated workspace/worktree is required, while a later section says no dedicated worktree is required. In an execution-oriented skill, this ambiguity can cause the agent to operate in the wrong repository state, including modifying a primary branch or shared workspace unexpectedly.

Vague Triggers

Medium
Confidence
89% confidence
Finding
The description is broad enough to match many normal implementation requests, which increases the chance this skill is invoked in situations the user did not explicitly intend. Because the skill then drives file reads, batch execution, and follow-on workflow behavior, overbroad triggering can cause unwanted autonomous changes.

Missing User Warnings

Medium
Confidence
94% confidence
Finding
The skill mandates updating multiple memory and project files such as progress and handoff artifacts, but it does not require obtaining explicit user consent before creating or modifying them. This can lead to silent persistence, repository noise, and unintended writes outside the user's requested task scope.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal