Back to skill

Security audit

Wiro AI Realtime Voice

Security checks for vulnerabilities and agentic risk

Overview

The skill is marked suspicious because it directs microphone audio to third-party services without a privacy notice and suggests an unpinned optional install; the scan evidence is incomplete.

Before installing, consider whether you are comfortable sending microphone audio to Wiro and the selected model provider, and check that the app provides clear notice and controls before recording. If you use the optional install command, review what it downloads and prefer pinned versions. The scan covered the listed Markdown component but was marked partial because two local references could not be resolved.

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
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% 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 · SKILL.md (reported line 23)May include surrounding context.

md
## Start a session

1. Pick a model and read its parameters with `POST /Tool/Detail`. They differ per model.
2. From server-side code (see Credentials in apps), create the task with `POST https://api.wiro.ai/v1/Run/{owner}/{model}`. Use this asynchronous Run; the `/sync` variant is for finite models.
3. Open `wss://socket.wiro.ai/v1` and register with `task_info`.
4. Wait for `task_stream_ready`. Send no audio before it.
5. Stream microphone frames, play the audio frames that come back, or both.

Microphone streaming is instructed without a user-facing privacy warning about third-party audio processing.

Medium
Category
Output Handling
Confidence
90% confidence
Finding

The skill instructs applications to stream microphone audio to Wiro and use external model providers without requiring a user-facing disclosure of that processing before capture starts. Although it requires microphone permission and appropriate session cleanup, those safeguards alone do not explain where potentially sensitive speech or bystander audio goes; the omission can propagate into applications built from this guidance.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx skills'.

Medium
Category
MCP Rug Pull
Confidence
86% confidence
Finding

The suggested npx skills add command runs an unpinned package and installs skill content from an unpinned repository reference, creating a supply-chain exposure if either source is compromised or changes unexpectedly. This is an optional dependency-installation recommendation, not an MCP server configuration as the static finding suggests, and there is no evidence of an actual compromise.

Content

No source excerpt is available for this finding.