ClawHub
Back to skill

Security audit

Taku Build

Security checks across malware telemetry and agentic risk

Overview

This is an instruction-only coding workflow skill that can change project files and run local setup commands, but its behavior is disclosed and aligned with implementing approved plans.

Install this only if you want an agent to implement approved coding plans. Expect it to edit files, run tests, install dependencies, and possibly create worktrees or a small .gitignore commit; review the preflight output and ask the agent to confirm before dependency installs or commits in sensitive repositories.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • 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 (3)

Vague Triggers

Medium
Confidence
89% confidence
Finding
The skill advertises many generic trigger phrases such as 'build this', 'start coding', and multilingual equivalents, which can cause the build skill to activate in contexts where the user did not intend to authorize code changes. Because this skill is designed to execute implementation work, accidental invocation could lead to unintended file edits, branch/worktree creation, or autonomous coding actions.

Vague Triggers

Medium
Confidence
93% confidence
Finding
The trigger phrases are broad enough to activate this skill for common requests like creating an isolated environment or starting feature work, which can cause the agent to perform repository-affecting actions unexpectedly. In this skill, activation leads to git worktree creation, dependency installation, test execution, and possibly repository modification, so accidental invocation has meaningful side effects.

Missing User Warnings

Medium
Confidence
97% confidence
Finding
The skill instructs the agent to append to .gitignore and immediately create a commit without explicit user approval, which modifies repository history and can create unauthorized commits on the user's branch. This is especially risky because the action is framed as an automatic safety fix, making silent persistence of changes more likely during routine setup.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.