ClawHub
Back to skill

Security audit

SUPAH Pulse

Security checks across malware telemetry and agentic risk

Overview

SUPAH Pulse is a clearly disclosed paid crypto-market briefing skill, but users should control when it can spend USDC and where it writes results.

Install only if you are comfortable with a paid crypto-market data skill. Use a low-balance wallet or spending limits, require explicit confirmation before paid calls if your agent supports it, leave SUPAH_API_BASE unset unless you trust the alternate endpoint, and do not treat the market call as financial advice.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Description-Behavior Mismatch

Low
Confidence
88% confidence
Finding
The script writes output to a predictable fixed path in /tmp (/tmp/market-pulse-result.json). On multi-user systems this can enable clobbering, symlink attacks, or unintended disclosure of generated results to other local processes, especially if the script is run with elevated privileges or by shared automation.

Vague Triggers

Medium
Confidence
89% confidence
Finding
The invocation phrases are extremely broad and overlap with ordinary market-related conversation, making accidental activation plausible. In this skill, accidental activation is more serious because each invocation may trigger a paid network call, so routine chat about markets could cause unintended spending and external data access.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill states that the agent pays automatically per call, but frames this as product behavior rather than a clear user-facing warning about spending. Because the skill can be triggered by broad natural-language prompts, users may incur repeated micropayments without informed consent, creating a real risk of unintended financial loss.

Missing User Warnings

Low
Confidence
90% confidence
Finding
The script silently persists normal-execution results to a fixed file in /tmp without explicit consent or a configurable destination. This creates the same local confidentiality/integrity risks as other predictable temporary-file usage and may surprise callers that expect stdout-only behavior from a briefing utility.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.