Back to skill

Security audit

上下文记忆球

Security checks for vulnerabilities and agentic risk

Overview

This is a Markdown-only context snapshot skill whose save/load behavior matches its stated purpose, but users should treat saved memory balls as potentially sensitive persisted data.

Install only if you want an agent to manage reusable conversation snapshots. Before saving or loading a memory ball, confirm what will be stored, where it will be kept, whether it contains private information, and whether automatic archive or deletion behavior is enabled.

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
  • 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 (5)

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
82% confidence
Finding

The skill description, instructions, and activation examples are entirely presented in Chinese, and no alternative language or user choice is offered. For a general-purpose skill, this effectively forces a specific language/locale without opt-in or documented justification.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The activation phrases include very generic requests such as '保存上下文' and '加载上下文', which are likely to overlap with ordinary user conversation. That can cause the skill to trigger unintentionally and save or restore conversation state without sufficiently explicit user intent, creating privacy and integrity risks around context handling.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill describes saving to files and exporting JSON but does not clearly warn users that conversation context may be persisted outside the current session. Without an explicit warning, users may unknowingly store sensitive data and later lose control over its retention or exposure.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The example '继续上次的工作' is an ambiguous continuation request that could automatically load the most recent active memory ball. This is dangerous because a routine conversational phrase may restore stale, unrelated, or sensitive prior context into the current session without clear user confirmation.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The cleanup policy includes automatic deletion of archived memory balls after 90 days, but the skill does not present this as an explicit user warning. This can lead to unexpected data loss and may also affect auditability or recovery if users assume archived context is retained indefinitely.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.