Security audit

Call My Agent

Security checks across malware telemetry and agentic risk

Overview

This is a coherent local voice-agent app, but it exposes powerful OpenAI and OpenClaw actions through web endpoints that are not clearly authenticated beyond the private network boundary.

Install only if you will keep this app reachable only by trusted devices and users. Keep the default localhost binding, use Tailscale rather than a public tunnel, set a strong CALL_MY_AGENT_CALL_TOKEN before remote use, and understand that anyone who can reach the app may be able to start voice sessions using your OpenAI access, invoke your OpenClaw agent, schedule reminders, or read/write this app's local capture state.

SkillSpector

By NVIDIA

Vulnerability Patterns

Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep

VirusTotal

59/59 vendors flagged this plugin as clean.

View on VirusTotal

Static analysis

Detected: suspicious.env_credential_access, suspicious.exposed_secret_literal

Environment variable access combined with network send.

Critical

Code: suspicious.env_credential_access
Location: app/api/session/route.ts:10
Evidence: const REALTIME_MODEL = process.env.OPENAI_REALTIME_MODEL ?? "gpt-realtime";

File appears to expose a hardcoded API secret or token.

Critical

Code: suspicious.exposed_secret_literal
Location: app/page.tsx:324
Evidence: await session.connect({ apiKey: [REDACTED] });