Back to skill

Security audit

research-summary

Security checks for vulnerabilities and agentic risk

Overview

This skill is a narrow Markdown-only helper for turning analyst or industry reports into Chinese investment-style summaries, with no hidden execution or persistence behavior found.

Install this if you want Chinese investment-style summaries of analyst or industry reports. Be aware that generic triggers may activate it in nearby finance conversations, and it may answer in Chinese unless you explicitly ask otherwise.

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
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (4)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The trigger 'analyst' is broad enough to match many ordinary finance or workplace queries unrelated to this skill's narrow purpose. This can cause unintended activation, leading the assistant to apply domain-specific formatting and assumptions to unrelated user requests, which is a prompt-routing integrity issue.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The shorthand trigger '180k' is ambiguous and lacks contextual bounds, so it may collide with unrelated inputs such as salaries, quantities, product specs, or IDs. Ambiguous activation increases the chance of misrouting user requests into this skill, causing unwanted behavior and reducing trust in assistant intent handling.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

Forcing output to always be in Chinese without checking user preference can override user intent and create unsafe mismatches in multilingual contexts, especially if the input is compliance-sensitive or needs faithful English retention. While not a classic exploit, it is a robustness and policy-control weakness because the skill unconditionally constrains response language.

Content

No source excerpt is available for this finding.

Overly Broad Trigger

Low
Category
Trigger Abuse
Confidence
88% confidence
Finding

The trigger '研报' is very short and common in this domain, making accidental activation more likely whenever users casually mention research reports. In a skill-routing system, short generic triggers can hijack broader conversations and steer the assistant into a specialized mode the user did not explicitly request.

Content

No source excerpt is available for this finding.