Back to skill

Security audit

Raven Transfer

Security checks across malware telemetry and agentic risk

Overview

This is a real Raven payout skill with disclosed purpose, but it needs review because its workflow can execute money transfers without a mandatory separate user approval step.

Install only if you intentionally want an agent to assist with Raven NGN payouts. Use a dedicated least-privilege Raven key if available, keep implicit invocation disabled, require the agent to show recipient, amount, fee, total debit, and balance impact before any --confirm command, and exclude scripts/.state from backups or shared repositories.

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
Findings (2)

Missing User Warnings

Medium
Confidence
94% confidence
Finding
The command reference documents live money-movement operations (`transfer` and `transfer-merchant`) with concrete execution examples but does not prominently warn that these commands can debit real funds and change account state. In an agent skill context, this omission increases the risk of accidental execution by users or downstream agents who may treat the examples as safe routine commands, especially because the skill is explicitly wallet-aware and payout-capable.

Missing User Warnings

Medium
Confidence
94% confidence
Finding
This workflow instructs the agent to execute a real bank payout and only says to "execute transfer once," but it does not require an explicit user-facing warning that funds will be irreversibly moved. In a wallet-transfer skill handling NGN payouts, that omission is dangerous because an agent could carry out a destructive financial action based on incomplete or implicit consent, increasing the risk of unauthorized or accidental transfers.

VirusTotal

66/66 vendors flagged this skill 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
tests/contract-live.test.mjs:6

File appears to expose a hardcoded API secret or token.

Critical
Code
suspicious.exposed_secret_literal
Location
tests/unit-normalizers.test.mjs:171