Security audit

openclaw-whobot-skill

Security checks for vulnerabilities and agentic risk

Overview

This is a WhoBot product knowledge skill with a local MCP query server, and the reviewed files do not show hidden data access, credential use, persistence, or destructive behavior.

Install this if you want WhoBot-specific product answers. Prefer the default stdio MCP setup, do not expose the optional HTTP server to untrusted networks, and review any external GitHub sync/update workflow before running it. For real deployments involving call recordings, healthcare calls, CRM updates, or webhooks, handle consent, retention, access control, and regulatory review separately.

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 (2)

Vague Triggers

Medium

Confidence: 91% confidence
Finding: The trigger condition includes broad topic matching such as 'any WhoBot-related topics,' which can cause unintended activation on loosely related queries. Over-broad invocation can route users into a branded knowledge skill without clear consent, increasing the chance of irrelevant answers, prompt hijacking of conversation flow, or accidental disclosure of skill-scoped content.

Natural-Language Policy Violations

Medium

Confidence: 81% confidence
Finding: The instruction to respond primarily in Chinese imposes a language preference regardless of the user's chosen language. While not a classic security bug, it can degrade user control and make the agent behave unexpectedly, which is risky in multi-user or multilingual contexts where clarity and consent matter.

Static analysis

No suspicious patterns detected.