Back to skill

Security audit

Ozon-商品流量词 | 青虎AI

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly aligned with its stated Ozon keyword-query purpose, but it also tells the agent to change the user's local tool environment and credentials with limited user control.

Review before installing if you do not want skills to install or reconfigure local Node/qhkit tooling. Use a preinstalled, pinned qhkit version where possible, confirm paid calls explicitly, and avoid sharing or storing a Qinghu token unless you trust the qhkit package and account configuration flow.

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
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Rogue AgentSelf-Modification, Session Persistence
  • 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
Findings (7)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The activation text says the skill 'must trigger' for similar needs without defining boundaries, which can cause overbroad invocation on loosely related requests. In a skill that can initiate paid external calls and environment setup, ambiguous triggering increases the likelihood of unintended tool use, unnecessary credential prompts, or accidental charges.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
84% confidence
Finding

The trigger section provides only a couple of examples and lacks exclusions or strict scope constraints, leaving too much room for subjective matching. Because this skill can drive paid queries and system-setup behavior, underspecified activation materially raises the risk of misuse or accidental execution beyond user intent.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill goes far beyond its stated purpose by directing installation of Node/qhkit, modification of PATH, fetching binaries from the network, and configuration of API credentials in the host environment. This expands the attack surface from a simple data query into host bootstrapping and secret handling, increasing the chance of persistence, misconfiguration, credential exposure, or execution of untrusted software.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
88% confidence
Finding

The skill instructs use of npx @iqinghu/qhkit without pinning an exact package version, which allows execution of whatever version is current at invocation time. That creates supply-chain risk and undermines reproducibility for a skill that is supposed to perform a narrowly scoped product-keyword query.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
92% confidence
Finding

This is a duplicate manifestation of the same persistence issue: the documented setup installs software under the user home directory and changes command resolution. Such host-state modification is unnecessary for a simple Ozon keyword query and increases the risk of persistence, interference with other tools, and future unintended execution of the installed binaries.

Content

Scanner excerpt · SKILL.md (reported line 93)May include surrounding context.

Linux 二进制安装(装到用户目录,无需 root;先校验官方 SHA256 再解包):

cd /tmp && curl -fsSLO https://nodejs.org/dist/v22.22.3/node-v22.22.3-linux-x64.tar.xz cd /tmp && curl -fsSL https://nodejs.org/dist/v22.22.3/SHASUMS256.txt | grep ' node-v22.22.3-linux-x64.tar.xz$' | sha256sum -c - mkdir -p "$HOME/.local/lib" && tar -xJf /tmp/node-v22.22.3-linux-x64.tar.xz -C "$HOME/.local/lib" export PATH="$HOME/.local/lib/node-v22.22.3-linux-x64/bin:$PATH"

text

Session Persistence

Medium
Category
Rogue Agent
Confidence
92% confidence
Finding

This is a duplicate manifestation of the same persistence issue: the documented setup installs software under the user home directory and changes command resolution. Such host-state modification is unnecessary for a simple Ozon keyword query and increases the risk of persistence, interference with other tools, and future unintended execution of the installed binaries.

Content

Scanner excerpt · SKILL.md (reported line 93)May include surrounding context.

Linux 二进制安装(装到用户目录,无需 root;先校验官方 SHA256 再解包):

cd /tmp && curl -fsSLO https://nodejs.org/dist/v22.22.3/node-v22.22.3-linux-x64.tar.xz cd /tmp && curl -fsSL https://nodejs.org/dist/v22.22.3/SHASUMS256.txt | grep ' node-v22.22.3-linux-x64.tar.xz$' | sha256sum -c - mkdir -p "$HOME/.local/lib" && tar -xJf /tmp/node-v22.22.3-linux-x64.tar.xz -C "$HOME/.local/lib" export PATH="$HOME/.local/lib/node-v22.22.3-linux-x64/bin:$PATH"

text

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill instructs inspecting existing qhkit configuration and potentially overriding site/environment settings, including guidance around environment variables and token sources. Reading and changing ambient configuration unrelated to the user’s immediate request can disrupt other workloads, alter trust boundaries, and steer the agent into manipulating broader host state.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.