ClawHub
Back to skill

Security audit

reCamera Intellisense

Security checks across malware telemetry and agentic risk

Overview

This skill is a disclosed reCamera control tool, with sensitive camera, storage, GPIO, and credential behavior that fits its stated purpose.

Install only for cameras you own or administer. Treat saved reCamera tokens and relay URLs as sensitive, review commands before allowing media capture or GPIO/storage changes, and require explicit user intent before passing confirm=true for destructive operations.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (5)

Lp3

Medium
Category
MCP Least Privilege
Confidence
92% confidence
Finding
The skill exposes sensitive capabilities including network access, filesystem reads/writes, environment-variable use, and shell invocation, but does not declare equivalent permissions in a structured way. That gap weakens review and consent boundaries, making it easier for an agent or operator to invoke a skill with broader effects than expected, including access to stored tokens and device control.

Tp4

High
Category
MCP Tool Poisoning
Confidence
82% confidence
Finding
The top-level description emphasizes camera onboarding, detection, polling, capture, and workflow integration, but the command set also includes broader administrative and operational controls such as rebooting devices, modifying ISP/image settings, and inspecting device/system state. This mismatch can mislead users or calling agents about the skill's real authority, increasing the chance of unintended disruptive actions on physical devices.

Vague Triggers

Medium
Confidence
88% confidence
Finding
The trigger guidance says to invoke the skill whenever the user mentions reCamera or a registered reCamera device, including broad workflow-related requests in that context. This is overly permissive for a skill that can write files, manipulate GPIO, clear event buffers, reboot hardware, and manage storage, creating a risk of accidental activation from ordinary conversation and unintended execution against real devices.

Missing User Warnings

Medium
Confidence
85% confidence
Finding
The local auto-detection path reads RECAMERA_TOKEN from the environment and then silently persists it to ~/.recamera/devices.json for future reuse. In an agent skill that can be triggered automatically in reCamera-related workflows, this expands the lifetime and exposure surface of a sensitive token beyond the user's ephemeral environment, increasing risk from local file disclosure, backups, or later unintended access.

Missing User Warnings

Medium
Confidence
89% confidence
Finding
The function activates an HTTP-triggered recording event immediately via a POST request, which can cause image/video capture on a physical camera without any confirmation, consent check, or operator-visible warning in this code path. In the context of a camera-control skill, this is security-relevant because an agent could be induced to perform real-world surveillance actions from ambiguous or malicious prompts.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

Detected: suspicious.insecure_tls_verification

HTTPS certificate verification is disabled.

Warn
Code
suspicious.insecure_tls_verification
Location
scripts/recamera_intellisense/_http.py:19