Security audit

Kami Smarthome Suite

Security checks across malware telemetry and agentic risk

Overview

This smart-home installer is mostly purpose-aligned, but it handles sensitive camera and notification credentials with under-scoped disclosure and has an installer that automatically runs the user's shell startup file.

Review before installing. This skill is not judged malicious, but it should be treated as a privacy-sensitive smart-home installer: run setup only when you explicitly intend to install the full Kami suite, avoid using credential-bearing RTSP URLs when possible, do not paste --show output into chats or logs, and be aware that setup.sh will execute your ~/.bashrc and install additional dependencies and skills.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands

Findings (6)

Context-Inappropriate Capability

Medium

Confidence: 98% confidence
Finding: The installer sources the user's full ~/.bashrc before performing setup, which executes arbitrary shell code from a user-controlled startup file in the installer's process. Because the script later performs network installs, package installation, and may invoke sudo paths, this creates a dangerous code-execution surface and makes the installer behavior depend on untrusted shell initialization.

Vague Triggers

Medium

Confidence: 84% confidence
Finding: Several triggers such as 'smart home suite' and 'kami home assistant' are broad enough to match ordinary user requests unrelated to installing this specific bundle. Overbroad invocation increases the chance the skill is activated unintentionally and then performs installation, configuration, or networked setup actions the user did not specifically request.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The setup flow asks users to enter API keys, RTSP URLs, and notification secrets and writes them into local configuration files, but the guidance does not prominently warn that these secrets will be stored on disk. Because RTSP URLs often embed camera usernames/passwords and notification tokens enable message delivery, local compromise or accidental file exposure could leak sensitive home surveillance and messaging credentials.

Missing User Warnings

Medium

Confidence: 93% confidence
Finding: The interactive template workflow prints the fully rendered RTSP URL, which typically embeds camera usernames and passwords directly in the string. This can expose credentials in terminal scrollback, logs, screen recordings, or remote shell transcripts, especially in an agent-driven or shared-admin environment.

Missing User Warnings

Medium

Confidence: 97% confidence
Finding: The --show command prints the centralized configuration while masking only the API key, leaving RTSP URLs, webhook URLs, bot tokens, app secrets, and chat/channel identifiers visible. Because this command is intended for inspection, it creates a straightforward secret disclosure path via console output, logs, or agent transcripts.

Missing User Warnings

Medium

Confidence: 97% confidence
Finding: Sourcing ~/.bashrc without an explicit warning or consent causes execution of arbitrary startup commands that the installer did not author or audit. Even if intended to initialize conda, this silently runs whatever is in the user's shell profile and can trigger unexpected commands, data exposure, or environment manipulation.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.