Security audit

Jupiter Prediction Market

Security checks across malware telemetry and agentic risk

Overview

This appears to be a legitimate Jupiter prediction-market client, but it needs Review because it enables automated financial actions without strong default safeguards.

Install only if you intentionally want an agent or script to interact with Jupiter prediction-market trading workflows. Prefer JUPITER_API_KEY from a secret store, avoid committing config/api-key.json, run claim workflows with --dry-run first, require manual review for wallet signatures and orders, and set explicit limits for order size, total exposure, and bulk position changes.

SkillSpector

By NVIDIA

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
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

Findings (6)

Lp3

Medium

Category: MCP Least Privilege
Confidence: 89% confidence
Finding: The skill advertises environment-variable use and external API access but does not declare permissions or clearly bound those capabilities. In an agent setting, undeclared env and network access reduce transparency and can cause the host to grant broader access than users expect, increasing the chance of secret exposure or unintended external actions.

Intent-Code Divergence

Medium

Confidence: 99% confidence
Finding: The `get(positionPubkey)` method shadows the inherited HTTP `get` method and then calls `this.get(...)`, which recursively invokes itself instead of issuing a network request. This causes denial of service via stack overflow or makes position retrieval unusable, and in a trading agent context can break monitoring or settlement workflows at runtime.

Missing User Warnings

Medium

Confidence: 93% confidence
Finding: The skill repeatedly promotes autonomous trading, portfolio management, and execution workflows without prominent warnings that these actions can trigger real financial transactions, losses, and irreversible wallet-linked operations. In this context, omission of risk and consent messaging is dangerous because users may deploy the skill as an autonomous agent and underestimate the consequences of live trading behavior.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The documented auto-claim workflow presents a wallet-linked asset action as a routine script invocation without a strong warning that it can submit real claims affecting user assets. Even if a wallet signature is required, describing batch claiming and automation without explicit consent and dry-run-first guidance can lead to unintended live operations in an agentic environment.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: The example shows `client.claim(...)` executing payout-claim transactions in a loop without any warning that it changes blockchain/account state or may incur fees, fail partially, or require user authorization safeguards. In the context of an agent skill for autonomous trading and portfolio management, this omission is more dangerous because developers may copy the snippet into automation that triggers real on-chain actions against user wallets.

Missing User Warnings

Medium

Confidence: 93% confidence
Finding: The auto-claim workflow performs real account-affecting actions by iterating over all claimable positions and calling `client.claim(...)` without any confirmation, guardrails, rate limiting, or explicit warning in the surrounding documentation. In an autonomous agent context, this increases the chance of unintended on-chain transactions, fee expenditure, or operational mistakes if the wrong wallet address/environment is used.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.