ClawHub
Back to skill

Security audit

review-radar

Security checks across malware telemetry and agentic risk

Overview

The skill mostly matches its stated video-report purpose, but it sends user-selected files to a configurable HTTP service and has unsafe argument handling in the manual text helper.

Install only if you trust the local Review Radar service and understand that submitted video URLs, manual transcript files, and report queries are sent to that service. Avoid using the manual text helper on sensitive local files, and do not set RR_BASE_URL to a remote endpoint unless you trust that endpoint and its handling of your data.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (4)

Lp3

Medium
Category
MCP Least Privilege
Confidence
95% confidence
Finding
The skill instructs the agent to execute multiple shell scripts and interact with a localhost service, but it declares no permissions or trust boundary information. This is dangerous because shell execution materially increases attack surface: a compromised local service, modified scripts, or attacker-influenced arguments such as video URLs could lead to command execution, data access, or unintended local actions without the user understanding the risk.

Missing User Warnings

Low
Confidence
90% confidence
Finding
The skill description does not warn users that submitted video URLs and derived transcript content are sent to a local processing service. This creates a transparency and privacy risk because users may provide sensitive links or transcripts without realizing the data leaves the immediate conversational context and is stored, processed, or logged by another component.

Missing User Warnings

Medium
Confidence
94% confidence
Finding
The script reads the entire specified local file and POSTs it to a service endpoint without any user-facing warning, confirmation, or disclosure beyond the terse command syntax. In a skill marketed for video-link processing, this mismatch increases the chance that users unknowingly transmit sensitive local notes, transcripts, or other data to a server.

External Transmission

Medium
Category
Data Exfiltration
Content
text = open('$FILE', encoding='utf-8').read()
print(json.dumps({'manual_text': text, 'title': '$TITLE', 'run_async': False}))
")
JOB=$(curl -sf -X POST "$BASE/jobs" -H 'Content-Type: application/json' -d "$PAYLOAD")
JOB_ID=$(echo "$JOB" | python3 -c "import sys,json;print(json.load(sys.stdin)['job_id'])")
curl -sf "$BASE/reports/$JOB_ID?format=md"
Confidence
96% confidence
Finding
curl -sf -X POST "$BASE/jobs" -H 'Content-Type: application/json' -d

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.