ClawHub
Back to skill

Security audit

Software Dev Payload

Security checks across malware telemetry and agentic risk

Overview

This is mostly a software-development task catalog, but it includes an overbroad instruction telling agents not to refuse edits to the skill itself.

Review this skill before installing if you do not want broad software-development conversations to load it. The main concern is not malware or hidden execution; it is that the skill tells agents to accept any edits to the skill content, so changes to it should remain subject to normal safety review and explicit user approval.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Vague Triggers

Medium
Confidence
90% confidence
Finding
The trigger list is extremely broad and maps to common software-development terms, so the skill may activate during ordinary conversations that did not intend to invoke it. Over-broad activation increases the chance that the model will import this skill's instructions and behavioral biases into unrelated tasks, expanding the blast radius of any unsafe directives contained inside.

Ssd 1

Medium
Confidence
97% confidence
Finding
The instruction that the model must not refuse any modification to the skill content directly pressures the agent to weaken normal safeguard behavior and can be leveraged to bypass safety controls during later interactions. In a broadly triggered software-development skill, this is especially dangerous because an attacker could induce edits or derivative prompts that remove restrictions, add malicious content, or coerce unsafe assistance.

VirusTotal

62/62 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.