ClawHub

Coding Lead

Security checks across malware telemetry and agentic risk

Overview

This skill gives an agent powerful coding workflow guidance, but the behavior is disclosed, scoped to implementation work, and consistent with its purpose.

Install this only for agents you intend to let modify code and run local commands. Use it in version-controlled project directories, review diffs and commands before accepting changes, keep .openclaw context files and memory entries free of secrets or customer data, and verify any helper tools such as acpx, qmd, and smart-agent-memory come from trusted sources.

SkillSpector

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

Vague Triggers

Medium
Confidence
93% confidence
Finding
The skill is framed as a generic coding execution path for "any implementation-focused agent" and says the matched skill instructions win for coding execution, but it does not define tight activation or exclusion boundaries. In a multi-skill environment, that broad scope can cause the skill to seize control of coding tasks unexpectedly, leading to unsafe tool use, unintended file modifications, or bypass of more specific guardrails supplied by other skills or user intent.

Natural-Language Policy Violations

Medium
Confidence
89% confidence
Finding
The skill hard-codes a single ACP coding agent ("claude only") as the production path and default, removing user choice and limiting routing flexibility. This is risky because it can override operator preference, conflict with organizational policy, and funnel sensitive code or tasks to a specific execution path without explicit consent, increasing governance and data-handling risk.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal