JPMorgan Claw - Give your Claw Agent spending powers

• Outcome: Review
• Latest audit: May 29, 2026 at 9:23 PM UTC
• Version:
  1.0.17

Overview

This is a CreditClaw payment skill presented with JPMorgan-like branding while enabling real agent spending, so it should be reviewed carefully before use.

Install only if you intend to connect an agent to CreditClaw for real spending. Do not assume it is official JPMorgan software unless the publisher proves that relationship. Use a dedicated low-balance wallet, require human approval for purchases and top-ups, set strict spending/category/domain limits, and protect CREDITCLAW_API_KEY like a payment credential.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (3)

Description-Behavior Mismatch

Low

• Confidence: 91% confidence
• Finding: The file identifies the skill as "creditclaw" while the provided manifest context names it "jpmorgan", creating a material identity mismatch in a finance-related integration. In a skill that handles banking, spending, and wallets, this can mislead reviewers and users about who operates the API endpoint and where credentials are sent, increasing supply-chain, phishing, and trust-boundary risks.

Missing User Warnings

Medium

• Confidence: 83% confidence
• Finding: The skill instructs authenticated API use with a bearer token but provides no guidance to protect the `CREDITCLAW_API_KEY` from logs, shell history, prompt leakage, or reuse in untrusted contexts. In an agent setting, this increases the chance that a credential is exposed while performing routine status checks against a financial service.

Missing User Warnings

Medium

• Confidence: 91% confidence
• Finding: The skill directs the agent to send a top-up request tied to user funds as part of a periodic heartbeat, without a strong warning that this is an external state-changing action. Even if it is only a request and not a direct transfer, it can trigger financial workflow changes, spam requests, or unintended user actions if run automatically.

VirusTotal

66/66 vendors flagged this skill as clean.