ClawHub
Back to skill

Security audit

Outbound Call

Security checks for vulnerabilities and agentic risk

Overview

This skill is a straightforward outbound phone-call helper, but users should treat calls and call context as real third-party actions requiring explicit confirmation.

Install only if you intend agents to place real outbound phone calls through your ElevenLabs/Twilio setup. Use a limited ElevenLabs key if possible, require explicit user confirmation of the normalized phone number before every call, and avoid passing secrets or unnecessary personal data in the optional message or context fields.

A.I.G

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Taint TrackingDirect Taint Flow, Variable-Mediated Taint Flow, Credential Exfiltration Chain
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
Findings (8)

Tainted flow: 'req' from os.environ.get (line 54, credential/environment) → urllib.request.urlopen (network output)

Critical
Category
Data Flow
Confidence
90% confidence
Finding

Credentials or environment variables flow to a network sink. This is a high-confidence indicator of credential exfiltration.

Content

Scanner excerpt · call.py (reported line 60)May include surrounding context.

python
})

    try:
        with urllib.request.urlopen(req) as resp:
            result = json.loads(resp.read())
            # ElevenLabs may return HTTP 200 with an error in the body
            if not result.get("success", True) or "error" in result:

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · call.py (reported line 26)May include surrounding context.

python
context: Optional context to pass as a dynamic variable.
    """
    if not API_KEY:
        return {"error": "ELEVENLABS_API_KEY not set. Add it to your .env file."}
    if not AGENT_ID:
        return {"error": "ELEVENLABS_AGENT_ID not set. Add it to your .env file."}
    if not PHONE_NUMBER_ID:

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · call.py (reported line 28)May include surrounding context.

python
context: Optional context to pass as a dynamic variable.
    """
    if not API_KEY:
        return {"error": "ELEVENLABS_API_KEY not set. Add it to your .env file."}
    if not AGENT_ID:
        return {"error": "ELEVENLABS_AGENT_ID not set. Add it to your .env file."}
    if not PHONE_NUMBER_ID:

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · call.py (reported line 30)May include surrounding context.

python
context: Optional context to pass as a dynamic variable.
    """
    if not API_KEY:
        return {"error": "ELEVENLABS_API_KEY not set. Add it to your .env file."}
    if not AGENT_ID:
        return {"error": "ELEVENLABS_AGENT_ID not set. Add it to your .env file."}
    if not PHONE_NUMBER_ID:

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
96% confidence
Finding

The skill can access environment secrets and initiate network-backed outbound phone calls, but it declares no explicit tool scope or permission boundaries. In an agent ecosystem, this increases the chance the skill is invoked with more capability than intended, making unauthorized calls or secret misuse harder to constrain and audit.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The invocation phrases are broad enough to match ordinary conversation like 'call someone' or 'ring someone,' which can cause accidental or premature activation of a real-world action skill. Because this skill places outbound phone calls, overbroad triggering is more dangerous than in read-only skills: a false trigger can lead to unintended contact, charges, privacy issues, or social-engineering opportunities.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · call.py (reported line 14)May include surrounding context.

python
API_KEY = os.environ.get("ELEVENLABS_API_KEY", "")
PHONE_NUMBER_ID = os.environ.get("ELEVENLABS_PHONE_NUMBER_ID", "")

API_URL = "https://api.elevenlabs.io/v1/convai/twilio/outbound-call"


def make_call(to_number: str, first_message: str = "", context: str = "") -> dict:

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The skill sends a phone number and optional free-form context to a third-party API that initiates a real phone call, but the code itself provides no consent prompt, disclosure, or restriction on what may be placed in the context field. In an agent-skill setting, this increases privacy and misuse risk because sensitive user data could be relayed off-platform or spoken during a live call without sufficient user awareness.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.