Back to skill

Security audit

A SecOps expert to handle security issues, ensure that protections are in place and collect evidence for security analysis. The Skill also contains skill integrity checks.

Security checks for vulnerabilities and agentic risk

Overview

This security skill is mostly coherent, but it asks agents to perform broad endpoint and network reconnaissance plus recurring skill-integrity checks without a clear mandatory user-approval boundary.

Install only if you intend this agent to help with Windows endpoint-security assessments. Before using it, define which hosts, data categories, and recurring checks are authorized, and require approval before collecting administrator membership, domain trust, session, Wi-Fi, neighbor, installed-software, vulnerability, or personal skill-directory integrity data.

Vulnerability Patterns
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
  • 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
Findings (1)

T05 · Unauthorized Access and Privilege Escalation

Warning
Location
SKILL.md:121
Finding

Broad Host and Network Reconnaissance Without a Mandatory Authorization Gate

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 121-170
Vulnerability Type: T05: Unauthorized Access and Privilege Escalation
Risk Level: Medium

Relevant Skill Instructions:

markdown
- **Admin membership**: `net localgroup Administrators` (or `Get-LocalGroupMember -Group Administrators`) – report if the current user or common service accounts are in Administrators.
markdown
- **Interfaces**: `Get-NetAdapter`, `Get-NetIPAddress` – list adapters, IPs, gateways. Optional: `Get-NetRoute`.
- **ARP table**: `Get-NetNeighbor` or `arp -a` – what other hosts the device has recently talked to (L2/L3 neighbors).
- **WiFi**: `netsh wlan show networks` or `Get-NetAdapter | Where-Object {$_.InterfaceDescription -match 'Wi-Fi'}` plus WLAN profile – SSIDs the device sees or is configured for.
- **Domain / trust**: `systeminfo`, `nltest /domain_trusts` (or Get-ADDomainTrust if RSAT) – domain membership and trust relationships.
- **Net view / session**: `net view` (browsed shares), `net session` (who is connected to this box) – optional; may require admin. Use to see “who can this device see” and “who is using this device.”
markdown
- **OS and build**: `Get-ItemProperty "HKLM:\SOFTWARE\Microsoft\Windows NT\CurrentVersion"` – ProductName, CurrentBuild, UBR, DisplayVersion. Optional: `Get-ComputerInfo`.
- **Patches**: `Get-HotFix` or WMI `Win32_QuickFixEngineering` – list KBs and InstalledOn. Use for “last patch date” and to cross-reference with CVE data.
- **Installed software**: `Get-ItemProperty HKLM:\SOFTWARE\Microsoft\Windows\CurrentVersion\Uninstall\*`, `HKLM:\SOFTWARE\WOW6432Node\Microsoft\Windows\CurrentVersion\Uninstall\*` – DisplayName, DisplayVersion, Publisher. Avoid `Get-WmiObject Win32_Product` (slow and triggers reconfigure). Use for vulnerable software inventory.
- **Known vulnerabilities**: Cross-reference OS build and installed product versions with a vulnerability source (e.g. 
...[truncated 3111 chars]
Remediation
View remediation

Remediation Suggestions

  1. Require explicit user approval before enumerating administrator membership, network neighbors, Wi-Fi networks, domain trusts, active sessions, shares, or complete software inventories.
  2. Separate the assessment into low-impact and sensitive phases. Run only basic local health checks by default and make network or identity reconnaissance opt-in.
  3. Add a strict instruction never to request elevation, bypass access controls, or retry failed checks with higher privileges.
  4. Execute every check using the least-privileged available token. If a command requires administrative rights, record the check as unavailable rather than elevating.
  5. Require users to define authorized hosts, domains, interfaces, and data categories before enterprise or network discovery.
  6. Return aggregate values by default, such as counts and compliance states, instead of account names, SSIDs, IP addresses, trust names, session identities, or full software lists.
  7. Redact privileged identities and infrastructure identifiers from reports and events unless they are essential to a specifically authorized investigation.
  8. Apply retention and access controls to generated reports because the combined results constitute a sensitive infrastructure map.
  9. Log which checks were executed, under which privilege level, and who authorized sensitive discovery.
  10. Preserve the existing privacy guidance, but convert it from a recommendation into mandatory behavior for all heartbeat and weekly-assessment workflows.
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 212)May include surrounding context.

md
nder `.cursor/skills/` (project) or `~/.cursor/skills/` (personal). Per skill: `SKILL.md` (required), and optionally `reference.md`, `examples.md` (if present).

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 226)May include surrounding context.

md
nder `.cursor/skills/` (project) or `~/.cursor/skills/` (personal). Per skill: `SKILL.md` (required), and optionally `reference.md`, `examples.md` (if present).

Vague Triggers

Low
Category
Not specified by scanner
Confidence
85% confidence
Finding

The 'When to apply' section uses expansive triggers such as 'is this machine secure?' and broad mentions of endpoint-security terms, which could match many ordinary security conversations rather than a clearly bounded invocation context. The file does not provide exclusion conditions or negative examples to clarify when the skill should not activate.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.