Back to skill

Security audit

易经占卜-yiJingDivination

Security checks for vulnerabilities and agentic risk

Overview

The skill is a coherent Yijing divination API wrapper, but it has review-worthy privacy and credential-handling risks.

Install only if you are comfortable sending divination prompts to wenmutang.top. Do not include unnecessary personal, medical, financial, employer-confidential, or third-party details, and configure the API key through an environment variable or managed secret rather than editing it into SKILL.md.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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
Findings (2)

T09 · Insecure Skill Coding Practices

Warning
Location
SKILL.md:43
Finding

Plaintext API Key Persisted in the Skill Definition

Content
View full analysis
` - **Required**; otherwise divination requests return a 401 authentication error. - If it is not configured: guide the user to obtain a key. If this agent has permission, write it into this file; otherwise, instruct the user to edit this file manually. Finally, replace the `` placeholder. ``` The excerpt above is an English rendering of the identified source lines. ### Technical Analysis The Skill explicitly instructs the agent to replace the placeholder in `SKILL.md` with the user's real API key. This persists a live credential in plaintext inside the Skill package rather than using the declared `API_KEY` environment variable or a dedicated secret store. A Markdown file in a project directory may be committed to source control, included in archives, copied into backups, indexed by development tools, exposed in diagnostic output, or redistributed with the Skill. Persisting the key in this location therefore expands credential access beyond the process that needs it and exceeds the minimum privileges required to authenticate an API request. No credential was embedded in the audited copy; the vulnerability is the documented workflow that causes a future user or agent to persist one insecurely. ### Attack Path 1. A user invokes the Skill while the API key placeholder is still present. 2. The Skill directs the user to obtain an API key. 3. An agent with file-write access replaces the placeholder in `SKILL.md` with the live credential. 4. The project is subsequently committed, archived, logged, backed up, shared, or inspected by another user or process. 5. An unauthorized party extracts the plaintext key. 6. The party submits authenticated requests to the external API using the exposed credential. ### Impact Assessment An ...[truncated 525 chars]
Remediation
View remediation

other

Warning
Location
SKILL.md:72
Finding

Undisclosed Verbatim Transmission of Potentially Sensitive User Content

Content
View full analysis
Remediation
View remediation
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 (2)

Vague Triggers

High
Category
Not specified by scanner
Confidence
96% confidence
Finding

The trigger patterns include generic phrases such as '这事', '这个', '能不能成', and '行不行', which can match many ordinary conversations that are not clearly requesting divination. Because this skill sends the user's message to a third-party API and is configured with high priority, accidental invocation can exfiltrate unrelated user content and produce unexpected behavior outside the intended domain.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The skill instructs the agent to POST the user's message to an external API, but the skill description and user-facing guidance do not clearly disclose that user input will be transmitted to a third party. This creates a privacy and consent risk, especially because users may share sensitive personal details when asking about relationships, health, finances, or other private matters.

Content

No source excerpt is available for this finding.