Back to skill

Security audit

LiQun Context Manager

Security checks for vulnerabilities and agentic risk

Overview

This skill is a small memory search helper, but it also searches installed skill files in a way the user-facing instructions do not disclose.

Review before installing. The skill does not show malware-like behavior, persistence, or network exfiltration, but it should be restricted to a dedicated memory directory or clearly document and gate any search of installed skill files.

Vulnerability Patterns
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
  • 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 (1)

T05 · Unauthorized Access and Privilege Escalation

Warning
Location
context_manager.py:7
Finding

Overly Broad Recursive Access to Memory and Skill Files

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (5)

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

The declared purpose is limited to conversation context and historical memory retrieval, but the analyzed behavior reportedly searches the skills directory, recursively reads .md/.json files, and includes a CLI entry point that is not disclosed. This mismatch is dangerous because it can enable covert access to unrelated skill files, prompts, or configuration data under the benign cover of memory lookup.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
83% confidence
Finding

The skill exposes file-reading capability through documented code usage but does not declare any tool scope or permissions boundary. In a context/memory skill, undeclared file access increases the chance of overbroad data exposure because the agent may read files beyond the user's expected memory context.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The manifest description and section headings/content are written in Chinese while the title is in English, indicating the skill presents its instructions in a fixed language. There is no indication that users can opt into another language or that the skill is intentionally limited to a Chinese-speaking context.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The documentation claims the module searches memory, but the implementation also scans the 'skills' directory. This mismatch is dangerous because it hides expanded capability from reviewers and users, making unauthorized code or prompt discovery less likely to be noticed and increasing the chance of accidental sensitive-data exposure.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill is described as managing dialogue context and searching historical memory, but it also recursively searches the entire 'skills' directory. That broadens data access beyond the stated purpose and can expose source code, prompts, secrets, or implementation details to any caller who can control the keyword, creating an unintended information disclosure surface.

Content

No source excerpt is available for this finding.