ClawHub
Back to skill

Security audit

Mercado Libre Product Data API - Search, Products, Reviews, Questions, Best Sellers

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed Scavio API integration for Mercado Libre product data, with expected external requests and no hidden local persistence or privileged behavior.

Install only if you are comfortable sending Mercado Libre search terms, item IDs, listing URLs, and your Scavio API key to Scavio. Keep the API key in an environment variable or secret store, avoid pasting confidential/internal URLs as lookup input, and watch API credit usage when looping through pages.

A.I.G

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

SkillSpector

Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (5)

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
82% confidence
Finding

The instruction 'state the budget before looping' is embedded in an otherwise multilingual, region-specific skill, but the clearer policy issue appears in the overall document style and examples being exclusively in English while processing Spanish and Portuguese marketplace data. However, there is no explicit forced-language directive here, so this would be too speculative to flag as a policy violation.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
87% confidence
Finding

The example instructs making authenticated POST requests to an external API endpoint, transmitting both user query data and a bearer token. While expected for an API integration, this is still a genuine external transmission risk if users provide confidential search content or operators do not understand that data leaves the local environment.

Content

Scanner excerpt · SKILL.md (reported line 150)May include surrounding context.

Examples

bash
curl -X POST https://api.scavio.dev/api/v1/mercadolibre/search \
  -H "Authorization: Bearer $SCAVIO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"country": "mx", "query": "iphone 15", "free_shipping": true, "sort": "price_asc"}'

External Transmission

Medium
Category
Data Exfiltration
Confidence
87% confidence
Finding

The example instructs making authenticated POST requests to an external API endpoint, transmitting both user query data and a bearer token. While expected for an API integration, this is still a genuine external transmission risk if users provide confidential search content or operators do not understand that data leaves the local environment.

Content

Scanner excerpt · SKILL.md (reported line 150)May include surrounding context.

Examples

bash
curl -X POST https://api.scavio.dev/api/v1/mercadolibre/search \
  -H "Authorization: Bearer $SCAVIO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"country": "mx", "query": "iphone 15", "free_shipping": true, "sort": "price_asc"}'

External Transmission

Medium
Category
Data Exfiltration
Confidence
89% confidence
Finding

The skill sends user-supplied search terms, URLs, item IDs, and the bearer API credential to an external third-party service. This creates a real data egress boundary: sensitive user inputs or internal-only URLs pasted by a user could be transmitted off-platform, and the token is exposed to the remote service by design.

Content

Scanner excerpt · SKILL.md (reported line 166)May include surrounding context.

md
HEADERS = {"Authorization": f"Bearer {API_KEY}"}

def meli(endpoint, body):
    r = requests.post(f"{BASE}/api/v1/mercadolibre/{endpoint}", headers=HEADERS, json=body)
    r.raise_for_status()
    return r.json()["data"]

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
90% confidence
Finding

The prose says 'there is no question date' while the example object immediately includes answer.date. Although the distinction is likely intended to mean the question itself has no date field, the wording/documentation is contradictory enough to mislead users about what date information is actually returned.

Content

No source excerpt is available for this finding.