Back to skill

Security audit

IdentityMonitoringAgent

Security checks across malware telemetry and agentic risk

Overview

This skill is privacy-sensitive OSINT tooling, but its behavior is disclosed, user-directed, and aligned with its stated identity-monitoring purpose.

Install only if you understand that this runs third-party OSINT tools and sends searched emails, usernames, phone numbers, or other identifiers to external services. Use it only for your own identities or authorized investigations, expect rate limits and terms-of-service constraints, and run it in an isolated environment after reviewing the pip packages.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (7)

Context-Inappropriate Capability

Medium
Confidence
83% confidence
Finding
The skill description suggests passive web monitoring, but the implementation delegates core behavior to external reconnaissance binaries. This expands the trust boundary to locally installed third-party tools whose network behavior, logging, and output parsing are not controlled here, increasing supply-chain and privacy risk.

Missing User Warnings

Medium
Confidence
90% confidence
Finding
This skill is designed to probe email registration, username presence, and public leak references for personal identifiers, but it provides no warning about privacy, consent, legality, or misuse risks. In this context, the omission increases the chance of stalking, doxxing, or unauthorized enumeration because users are prompted to investigate third-party identities without any safeguards or limitations.

Missing User Warnings

Medium
Confidence
96% confidence
Finding
The function sends a potentially sensitive email address to an external OSINT tool without any notice, consent, or minimization. Even if intended for legitimate monitoring, this can expose personal identifiers to third-party services or logs and may violate user expectations or policy requirements.

Missing User Warnings

Medium
Confidence
96% confidence
Finding
The username is passed to an external reconnaissance tool with no warning that the value may be queried across multiple sites and services. This creates privacy and compliance risk because identifiers may be disclosed beyond the local environment without informed consent.

Missing User Warnings

Medium
Confidence
97% confidence
Finding
The search_leaks function sends arbitrary queries, potentially including sensitive identifiers or breach-related terms, to an external search provider without disclosure. This can leak investigative targets or private data to third parties and may leave audit trails outside the user's control.

Unvalidated Output Injection

High
Category
Output Handling
Content
def scan_email(email):
    try:
        result = subprocess.run(['holehe', email, '--only-used'], capture_output=True, text=True)
        raw_lines = result.stdout.split('\n')
        registered = [
            line.replace("[+]", "").strip()
Confidence
80% confidence
Finding
subprocess.run(['holehe', email, '--only-used'], capture_output

Unvalidated Output Injection

High
Category
Output Handling
Content
def scan_username(username):
    try:
        result = subprocess.run(['sherlock', username, '--timeout', '1'], capture_output=True, text=True)
        found = [line.split(': ')[1] for line in result.stdout.split('\n') if 'http' in line]
        return json.dumps({"target": username, "profiles": found})
    except Exception as e:
Confidence
82% confidence
Finding
subprocess.run(['sherlock', username, '--timeout', '1'], capture_output

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.