Token Burn Monitor

• Outcome: Pass
• Latest audit: May 29, 2026 at 8:18 PM UTC
• Version:
  5.3.1

Overview

This is a local token and cost dashboard with sensitive prompt display disclosed as an opt-in setting, so it is acceptable when kept local and configured carefully.

Install only on a machine where the dashboard may read OpenClaw session logs. Keep showPrompts and SHOW_PROMPTS disabled unless every dashboard viewer is trusted to see prompt text, keep the server bound to localhost, and review custom themes before enabling them because themes can display any data the API returns.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (3)

Description-Behavior Mismatch

Medium

• Confidence: 95% confidence
• Finding: The API returns per-message metadata including `userPrompt`, and prompt redaction is only conditional on `SHOW_PROMPTS` or `config.showPrompts`. That means a monitoring dashboard can expose sensitive user task content from agent session logs, which is broader access than the skill description suggests and increases risk of leaking secrets, internal instructions, or private requests to any local or network-accessible client.

Description-Behavior Mismatch

High

• Confidence: 99% confidence
• Finding: The UI explicitly renders `m.userPrompt` in an expandable row whenever a message includes a prompt, which contradicts the skill metadata claim that user prompts are redacted by default. In a token-monitoring dashboard, prompts may contain secrets, personal data, credentials, or sensitive operational context, so exposing them to any dashboard viewer materially increases data leakage risk.

Missing User Warnings

Medium

• Confidence: 90% confidence
• Finding: The API explicitly supports exposing `userPrompt` contents when `showPrompts` is enabled, but the documentation only notes the behavior and does not clearly warn that this can reveal potentially sensitive user inputs from session files. In a monitoring dashboard context, operators may enable the setting for convenience and inadvertently disclose secrets, personal data, or internal prompts to anyone with access to the dashboard or local API.

VirusTotal

66/66 vendors flagged this skill as clean.