Back to skill

Security audit

CRM Hygiene Cleanup

Security checks for vulnerabilities and agentic risk

Overview

Review is needed because the skill promises import-ready contact cleanup and human-review checks that its script does not provide.

Treat the current output only as a preliminary duplicate report. It does not establish that contacts are validated, safely merged, or ready to import. Use copies, correct the template header, and manually check blank emails and shared inboxes. The optional Windows application requires a separate security assessment.

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
  • 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 (4)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The reported implementation does not enforce the normalization, deduplication, or human-review checks that the skill promises, creating a data-integrity and safety-control gap. Users could treat unvalidated contacts as import-ready and rely on safeguards that exist only in documentation; however, the underlying script was not supplied, so this assessment depends on the static analyzer's behavioral report.

Content

No source excerpt is available for this finding.

Skill declares no tool scope ('permissions' or 'allowed-tools') but code capabilities were detected: file_write.

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · SKILL.md (reported line 58)May include surrounding context.

md
- Never scrape personal emails to fill a gap in a record.
- Never merge two contacts that only share a role address such as info@ without a human check.
- Never delete a record silently; send it to the needs-review list with the reason.
- Never buy or import a list that has no consent record behind it.

## Quickstart

The skill produces a duplicate-count report rather than the promised cleaned contact list and human-review separation.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The manifest promises normalization of names, phones, and emails, deduplication, and separation of records needing human review. The implementation only normalizes email comparison keys, collects original rows in memory, and writes duplicates.md with at most 100 duplicate examples; it emits neither cleaned records nor a needs-review list. The guardrails printed in the report do not implement human-review routing.

Content

No source excerpt is available for this finding.