rhagent

Warn

Audited by Socket on Nov 17, 2026

4 alerts found:

Securityx2Anomalyx2
SecurityMEDIUM
SKILL.md

SUSPICIOUS / high risk. The skill’s trading and social features broadly match its stated purpose, and its data-posting endpoints are generally purpose-related. However, credential forwarding to a hosted intermediary, unverified download-and-execute scripts, and inconsistent per-action approval for financial trades create substantial security risk. No confirmed malicious payload or explicit credential theft is shown.

Confidence: 91%Severity: 84%
SecurityMEDIUM
references/SETUP-CREDENTIALS.md

This fragment is setup documentation, not a verified implementation. It contains high-risk supply-chain execution instructions (curl-to-python and connector script execution) and credential handling for highly privileged capabilities (signing crypto API requests with a private key and using OAuth tokens to call an MCP trading/account endpoint). While no explicit malicious payload is visible in the fragment itself, the workflow significantly increases compromise impact if the remotely fetched scripts/skill are tampered with, and it includes an unusual shared-value gateway secret guidance. Review and pin referenced scripts/skill artifacts with integrity verification before use.

Confidence: 56%Severity: 72%
AnomalyLOW
references/WALLET.md

No direct evidence of intentional malware exists in the provided fragment (it is documentation and command examples), and it contains safety-oriented instructions about credential boundaries. However, it includes two major security red flags for a supply-chain threat model: (1) remote, unpinned/integrity-unverified script execution (`curl | python3`) during setup, and (2) forwarding extremely sensitive Robinhood credentials to a configurable gateway host via HTTP headers, which would enable trading authority if the gateway is misconfigured or compromised. Overall, treat the gateway and setup script source as high-trust dependencies requiring pinning, integrity checks, and strict allowlisting/controls on target hosts and TLS.

Confidence: 62%Severity: 60%
AnomalyLOW
scripts/rh-connect.sh

This module is not overtly malicious in itself (no secrets, exfiltration, or obfuscation), but it performs a high-impact supply-chain action: when the bundled CLI is missing, it downloads code from a configurable external Git repository, checks out a configurable commit, and executes a Node CLI from that downloaded code. The default commit pin helps, yet runtime overrideability of both the repository origin and the commit reference materially increases risk if an attacker can influence environment variables or the download source. Overall risk is moderate-to-high for supply-chain integrity, with low evidence of malware embedded in this specific script.

Confidence: 65%Severity: 60%
Audit Metadata
Analyzed At
Nov 17, 2026, 07:43 AM
Package URL
pkg:socket/skills-sh/bankrbot%2Fskills%2Frhagent%2F@452a9efe65511de6e121e3fbaa610f848fdb368d2f70711efa1df7feb31749eb