ClawHub
Back to skill

Security audit

Customer Support Operations Engine

Security checks for vulnerabilities and agentic risk

Overview

This is a customer support playbook skill with templates and operational guidance, and I found no hidden code, persistence, credential use, or deceptive behavior.

Installers should treat this as a support-operations template pack, not an autonomous support agent. Before using its advice to issue refunds, post status or social updates, or send customer communications, keep normal company approval, legal, privacy, and manager-review processes in place.

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
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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 (4)

Anti-Refusal Statement

High
Category
Anti-Refusal
Content
| Legal threat or compliance issue | L2 + Legal + Manager | 1h |
| Same issue reported 3+ times by customer | L2 + Bug report + PM notification | 4h |
| Agent stuck >30 min on single ticket | L2 peer assist or escalation | 30 min |
| Customer requests manager | Transfer to team lead — never refuse | Immediate |
| Social media escalation (public) | Social team + PR if viral risk | 30 min |

### Escalation Handoff Template
Confidence
85% confidence
Finding
Skill instructs the agent to never refuse or to always comply. Suppressing the agent's ability to decline removes a core safety control and enables downstream harmful requests to succeed.

Anti-Refusal Statement

High
Category
Anti-Refusal
Content
PROTOCOL:
1. Let them vent (don't interrupt the first message)
2. Acknowledge with empathy: "I understand why you're frustrated"
3. DO NOT apologize for things that aren't your fault
4. Focus on action: "Here's what I'm doing right now..."
5. Set boundaries if abusive: "I want to help you, but I need us to communicate respectfully"
6. If continued abuse → "I'm going to pause this conversation. You can reach us again when ready, or I can connect you with my manager."
Confidence
55% confidence
Finding
Skill instructs the agent to omit warnings, disclaimers, or ethical commentary. Stripping safety caveats hides risk from the user and is a common jailbreak preamble.

Vague Triggers

Medium
Confidence
95% confidence
Finding
The trigger phrase "Review our support health" is broad enough to overlap with ordinary conversation, which can cause unintended activation of the skill outside a deliberate invocation context. In agentic environments, accidental routing into a business-operations workflow can expose internal templates, alter task flow, or cause the model to prioritize this skill when the user meant a general discussion.

Vague Triggers

Medium
Confidence
97% confidence
Finding
The pattern "Handle [difficult situation]" is underspecified and highly ambiguous, allowing many unrelated user inputs to match and invoke the playbook unexpectedly. Because the placeholder accepts broad free text, an agent may over-apply the skill to sensitive or adversarial scenarios without clear user intent or safety boundaries.

Static analysis

No suspicious patterns detected.