Foxreach

Security checks across malware telemetry and agentic risk

Overview

Foxreach is a coherent FoxReach API helper, but users should handle its API key and live outreach actions carefully.

Install only if you trust the FoxReach SDK/CLI source and are comfortable giving an agent an API key that can manage outreach data. Prefer environment-based secrets, avoid pasting real keys into commands or saved snippets, and require an explicit preview before deletes, bulk changes, email-account removal, or starting campaigns.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (6)

Intent-Code Divergence

Medium

Confidence: 97% confidence
Finding: The skill explicitly says to never hardcode API keys, but its primary example uses a literal key string in source code. This normalizes unsafe secret handling and increases the chance that users or downstream agents will paste real credentials into scripts, shell history, logs, or saved files where they can be exposed.

Intent-Code Divergence

Medium

Confidence: 97% confidence
Finding: The one-liner embeds an API key literal directly in the command, contradicting the stated guidance to avoid hardcoding secrets. In practice, command-line secrets are especially risky because they may be captured in shell history, process listings, telemetry, or copied into chat/output logs.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: These examples demonstrate state-changing operations such as creating, updating, and deleting leads against what appears to be a live API, but provide no warning that running them will modify real customer/prospect data. In an outreach platform, this can cause accidental data loss, unauthorized contact record creation, or unintended changes to production CRM/outreach workflows by users who copy-paste examples verbatim.

Missing User Warnings

Medium

Confidence: 98% confidence
Finding: The full campaign setup example culminates in starting a campaign after adding leads and email accounts, without warning that this may trigger real outbound cold emails to actual recipients. Because this skill is specifically for managing outreach, users are especially likely to run examples operationally, creating immediate risk of unsolicited sending, reputational harm, spam complaints, and compliance issues.

Ssd 3

Medium

Confidence: 90% confidence
Finding: The examples print lead identifiers, email addresses, and statuses directly to output by default. In a cold-email platform context, this can expose personal/contact data in agent transcripts, logs, or shared terminals, creating unnecessary data disclosure risk.

Ssd 3

Medium

Confidence: 94% confidence
Finding: The auto-paging example encourages iterating over and printing all lead email addresses, which could dump an entire contact database into output. In this skill's context, that materially increases the risk of bulk PII exposure through transcripts, logs, or accidental sharing.

VirusTotal

55/55 vendors flagged this skill as clean.

View on VirusTotal