Back to skill

Security audit

Crm Data Cleaner

Security checks for vulnerabilities and agentic risk

Overview

This CRM-cleaning skill is purpose-aligned, but it asks for CRM credentials and includes broad bulk-update and third-party enrichment guidance that should be reviewed before use.

Install only if you are comfortable giving an agent access to CRM data. Start with least-privilege or read-only credentials, use exports or a sandbox first, require human approval before merges/imports/batch updates, keep backups, and verify legal/privacy approval before using Clearbit, Apollo, scraping, or social-data enrichment.

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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (6)

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The description promotes external enrichment providers and CRM update workflows but does not clearly warn that contact/company data may be sent to third parties or that privacy obligations may apply. Users may trigger enrichment assuming a local cleanup task, resulting in unintended disclosure of customer or prospect data to outside services.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The opening guidance describes deduplication, normalization, and enrichment workflows but lacks an immediate warning that merges and bulk updates can be irreversible or affect linked records, reports, and automation. In CRM systems, premature or automated merges can destroy auditability, overwrite accurate values, and disrupt downstream business processes.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The manifest explicitly claims the skill is instruction-only with no scripts or code execution, yet the document embeds multiple executable Python, JavaScript, SQL, and Apex snippets and references implementation workflows. This mismatch can mislead downstream tooling, reviewers, or users into granting the skill broader trust than warranted and increases the chance an agent copies or adapts code that performs external access or data mutation.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 254)May include surrounding context.

md
import requests

def find_hubspot_duplicates(api_key, batch_size=100):
    url = f"https://api.hubapi.com/contacts/v1/lists/all/contacts/all"
    params = {
        'hapikey': api_key,
        'count': batch_size,

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The skill recommends web scraping, social-media harvesting, and use of public databases to enrich CRM records, which can cause collection and transfer of personal data beyond what is necessary for basic deduplication/normalization. In a CRM context this creates privacy, consent, and terms-of-service risks, especially if an agent operationalizes scraping without legal review or user awareness.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
91% confidence
Finding

The inline note stating that implementation details exist in a scripts directory directly contradicts the no-scripts claim in the manifest. Even without the scripts present, this suggests hidden or omitted executable components and undermines trust in the declared security boundary of the skill.

Content

No source excerpt is available for this finding.