Back to skill

Security audit

Clawhub

Security checks for vulnerabilities and agentic risk

Overview

This skill matches its directory-submission purpose, but it needs review because it enables agents to send listing/contact data externally and documents autonomous USDC paid bids without clear confirmation safeguards.

Install only if you are comfortable with agents sending product URLs, owner emails, and listing metadata to CurlShip. Require explicit approval before any submit, edit, removal, upgrade, featured bid, or x402/USDC signing step, and verify the amount, URL, refund terms, and public-listing impact before proceeding.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • 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 (5)

External Transmission

Medium
Category
Data Exfiltration
Confidence
91% confidence
Finding

The skill is explicitly designed to transmit user-supplied data to an external service and to facilitate payment-related actions, including automated USDC payment flows. External transmission is expected in context, but it is still security-relevant because agents may disclose URLs, emails, listing metadata, and potentially payment authorization material to a third party without strong consent or trust gating.

Content

Scanner excerpt · SKILL.md (reported line 3)May include surrounding context.

md
---
name: CurlShip Directory Submission
description: Submit products to CurlShip, the bot-friendly directory. One curl command to list your product with OG tag scraping, badge-based dofollow links, and tier upgrades - paid by card checkout or by agents in USDC over x402 with no human in the loop.
version: 1.3.0
homepage: https://curlship.com
credentials_required: false

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill instructs users/agents to send a product URL and contact email to CurlShip, a third-party service, but it does not clearly warn about that data leaving the local environment. This can lead to unintentional disclosure of user-provided personal or business-sensitive information, especially when an autonomous agent submits data without explicit user confirmation.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
87% confidence
Finding

This endpoint triggers an external bid/payment flow for a featured listing, which could cause an agent to initiate financially meaningful actions against a third-party service. In an autonomous-agent setting, that creates risk of unauthorized spending or unintended business commitments if the user has not explicitly approved the action.

Content

Scanner excerpt · SKILL.md (reported line 171)May include surrounding context.

POST /api/feature — Bid on the spot. One-time payment; highest payment holds it.

bash
curl -X POST https://curlship.com/api/feature \
  -H "Content-Type: application/json" \
  -d '{"url": "https://yourapp.com", "amount_usd": 51}'

External Transmission

Medium
Category
Data Exfiltration
Confidence
85% confidence
Finding

The edit-request flow sends listing identifiers, URLs, and proof-selection data to an external service, which is a real outbound data transfer. While this is core functionality rather than malicious behavior, it can expose business information or trigger account-affecting operations on a third-party platform if agents act without sufficient user awareness.

Content

Scanner excerpt · SKILL.md (reported line 208)May include surrounding context.

bash
# Point the listing at a new URL
curl -X POST https://curlship.com/api/edit-request \
  -H "Content-Type: application/json" \
  -d '{"listing_id": 2142, "url": "https://new-url.example/", "proof": "email"}'

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
82% confidence
Finding

The text instructs users to 'Say 18+ in the product's own og:description', which imposes a specific labeling format in natural language without any opt-in or locale flexibility. This can conflict with language/locale policy expectations where users should be allowed a choice or the constraint should be explicitly justified.

Content

No source excerpt is available for this finding.