Back to skill

Security audit

Hong Kong Horse Racing

Security checks for vulnerabilities and agentic risk

Overview

The skill retrieves public horse-racing information and generates disclosed recommendations, with no evidence of hidden access or harmful actions.

Expect English output and experimental racing picks; verify current odds and results before relying on them. Use the supplied lockfile for reproducible dependency installation. All seven bundled files were inspected, including the scanner's partial-coverage files; external dependency code remains outside this review.

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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (5)

Skill declares no tool scope ('permissions' or 'allowed-tools') but code capabilities were detected: network.

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.

The skill mandates English-only output without user opt-in.

Medium
Category
Output Handling
Confidence
99% confidence
Finding

“English output only” imposes a specific response language without offering the user a choice. Although the skill reads English-language source pages, that does not justify requiring all responses to be in English.

Content

No source excerpt is available for this finding.

The output language is restricted to English without user opt-in.

Low
Category
Output Handling
Confidence
97% confidence
Finding

The language parameter accepts only "en", and its description explicitly requires English output by default. The manifest provides neither an alternative language choice nor a documented justification for this restriction.

Content

No source excerpt is available for this finding.

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
80% confidence
Finding

The caret range permits installation of newer hkjc-api releases without changing this manifest, creating a conditional supply-chain exposure if a permitted release is compromised. Dependency code runs within the Node.js process, so compromise could affect the host rather than merely the public racing data; however, no compromised version is identified, and a lockfile outside the supplied source could mitigate resolution drift.

Content

Scanner excerpt · package.json (reported line 10)May include surrounding context.

json
"author": "Steven Ho",
  "license": "MIT",
  "dependencies": {
    "hkjc-api": "^1.0.0"
  }
}

Recommendation reasons are explicitly fixed to English without a user language choice.

Low
Category
Output Handling
Confidence
94% confidence
Finding

The comment explicitly specifies “Reason phrases: fixed English,” and the phrase table implements that constraint for recommendation output. No language opt-in or justification for English-only explanations is provided; the Hong Kong racing context does not itself require English output.

Content

No source excerpt is available for this finding.