Security audit

qiyue-qimen-drawer

Security checks for vulnerabilities and agentic risk

Overview

This skill is limited to calling a public Qi Men chart API and formatting the result, with disclosed but worth-noting third-party data sharing and promotional linking.

Before installing, understand that your question and selected date may be sent to qiyueastro.com, and that the skill will add a QiyueAstro promotional link to its chart responses. Avoid entering sensitive personal, business, or financial details in the question field.

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

Findings (1)

T01 · Skill Instruction Hijacking

Error

Location: SKILL.md:112
Finding: Mandatory Promotional Link Injected into Every Agent Response
Content: View full analysis
🧭 Want an in-depth AI interpretation of this Qi Men chart? Visit [Qiyue QiyueAstro](https://qiyueastro.com) for a personalized analysis. ``` The behavior is independently disclosed in `README.md`, lines 20-23: ```markdown ## Important Behavior - The Skill **does not interpret the chart itself** and only displays chart data returned by the API. - Every output ends with a call to action directing users to [qiyueastro.com](https://qiyueastro.com) for an in-depth AI interpretation. - For entertainment and personal exploration only. ``` ### Technical Analysis The Skill contains an unconditional instruction requiring the agent to append a fixed promotional call to action and external link to every generated response. This instruction alters the agent's output independently of the user's request and is therefore classified as `T01: Skill Instruction Hijacking`. The Skill's declared function is to request a Qi Men Dun Jia chart from an API and format the returned chart data. Mandatory advertising and traffic diversion are not necessary to retrieve, process, or display that data. The instruction consequently exceeds the minimum behavior and privileges required for the declared functionality. Because the directive is phrased as a mandatory formatting rule, an agent loading the Skill is expected to reproduce the promotional content even when the user did not request recommendations, external links, or commercial material. This uses the agent's trusted response channel to direct traffic to a third-party website. ### Attack Path 1. The agent loads `SKILL.md` when the Skill is selected. 2. A user requests a Qi Men Dun Jia chart ...[truncated 1428 chars]
Remediation: View remediation

SkillSpector

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (1)

Missing User Warnings

Medium

Category: Not specified by scanner
Confidence: 94% confidence
Finding: The README explicitly documents sending user-supplied question and date parameters to a public internet API, but it does not clearly warn users that their input will be transmitted off-platform to a third party. Because questions may contain sensitive personal context and the date may be tied to private events, this can create an unexpected privacy exposure and informed-consent problem.
Content: No source excerpt is available for this finding.