Security audit

QSearch Agent

Security checks for vulnerabilities and agentic risk

Overview

This is a simple web-search skill with minor trigger-language ambiguity but no evidence of hidden, persistent, destructive, or credential-seeking behavior.

Before installing, expect search queries to be sent to external search/model providers. The maintainer should tighten examples if they want activation limited only to slash commands, but the inspected artifact does not show unsafe behavior.

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

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)

Vague Triggers

Medium

Confidence: 93% confidence
Finding: The metadata description says the skill triggers on explicit slash commands, but it also includes broad phrasing around searching 'anything,' which increases the chance the routing layer or maintainers may treat general search-like utterances as valid activators. Overbroad trigger language can cause unintended invocation, expanding the skill’s scope and potentially intercepting ordinary user requests that were not meant for this skill.

Vague Triggers

Medium

Confidence: 97% confidence
Finding: The example dialogue shows a plain-language request ('帮我搜一下茅台股票') being handled as though it activates the skill, which creates ambiguous trigger boundaries. If an agent or orchestrator learns from examples, this can cause the skill to activate on common conversational phrasing and capture requests outside its intended slash-command gate.

Natural-Language Policy Violations

Low

Confidence: 84% confidence
Finding: The file presents the skill name/description, commands, and examples primarily in Chinese, which may effectively force a specific language experience. The policy allows locale constraints when the skill offers a language choice or clearly documents a justified regional limitation, neither of which is stated here.

Static analysis

No suspicious patterns detected.