ClawHub
Back to skill

Security audit

Kiln

Security checks across malware telemetry and agentic risk

Overview

Kiln is a coherent 3D-printing skill, but it gives agents broad control over physical printers, cameras, and paid fulfillment, so users should review it carefully before enabling it.

Install only if you are comfortable letting an AI client control real 3D printers and related services. Keep preview and confirmation gates enabled, do first runs supervised, verify emergency-stop access, avoid unattended prints without fire precautions, set spend limits before fulfillment orders, and disable camera, cloud sync, and webhook features you do not need.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Rogue AgentSelf-Modification, Session Persistence
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Missing User Warnings

Medium
Confidence
97% confidence
Finding
The README prominently markets fully autonomous control of physical printers and outsourced ordering ('No human in the middle') without an equally prominent warning that actions can cause physical damage, fire risk, material waste, and financial spend. In an agent skill context, this framing encourages over-trust and makes unsafe autonomous execution more likely, especially when the agent is told it can design, slice, queue, monitor, recover, and ship end-to-end.

Missing User Warnings

Medium
Confidence
98% confidence
Finding
The quick-start tells users that the AI 'can run your printer' and 'does the rest' immediately after install, but does not place nearby warnings about machine motion, heaters, unattended operation, or external purchase/print actions. Because this is the operational onboarding section, the omission materially increases the chance that users enable unsafe autonomous real-world actions without guardrails.

Missing User Warnings

Medium
Confidence
89% confidence
Finding
The skill advertises camera snapshots and print monitoring but does not warn users that these features may capture sensitive visual data from homes, workshops, or surrounding environments. In a tool that enables direct AI-agent control of physical devices, omission of this privacy notice increases the risk of unintended collection, retention, or sharing of surveillance-like data.

Self-Modification

High
Category
Rogue Agent
Content
That's the whole happy path. Your AI does the designing, slicing, and printing — you never touch a CLI. OS-specific walkthrough (Windows, WSL 2, Linux) at **[kiln3d.com/install](https://kiln3d.com/install?utm_source=github&utm_medium=readme)**.

> [!TIP]
> `kiln doctor` checks your whole setup and tells you exactly what to fix. `kiln self-update` upgrades in place when a new release lands.

<details>
<summary><strong>Prefer to drive it yourself? (CLI tour)</strong></summary>
Confidence
71% confidence
Finding
Documenting an in-place self-update command for an agent-facing tool can be risky because autonomous agents may invoke it to change their own operational surface, potentially bypassing review, altering behavior, or introducing supply-chain risk during a session. In a physical-control skill, unsupervised runtime updates are more dangerous because they can modify tooling that directly affects printers, network access, and purchasing workflows.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.