Security audit

Watcha Finder

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed AI-product recommendation helper that uses watcha.cn and web searches, with no executable code, persistence, or local data access.

Install this if you want AI product recommendations informed by watcha.cn community data. For neutral or non-China-focused recommendations, ask the agent to compare watcha.cn results with global sources and to avoid using watcha.cn if you do not want those queries sent there.

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

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (3)

Vague Triggers

High

Confidence: 95% confidence
Finding: The skill’s activation criteria are so broad that it can trigger for generic AI-tool queries that may not specifically require watcha.cn. This creates an overreach risk: the agent may route users into a third-party platform workflow unnecessarily, shaping recommendations through a single data source and external web lookups without explicit user consent or clear necessity.

Natural-Language Policy Violations

Medium

Confidence: 88% confidence
Finding: The instruction set anchors the workflow to watcha.cn and explicitly directs the agent to use Chinese-platform data and Chinese-language search strategies by default, even when the user has not opted into China-focused sources. This can misalign results with user expectations, introduce regional bias, and potentially expose user queries to an external platform unnecessarily.

Natural-Language Policy Violations

Low

Confidence: 79% confidence
Finding: This JSON file contains natural-language prompts in both Chinese and English, but there is no indication that language or locale is user-selectable or intentionally constrained. Under the language/locale policy check, mixed or imposed language behavior can be a policy concern when no opt-in or justification is provided.

Static analysis

No suspicious patterns detected.