Back to skill

Security audit

Skill Memory

Security checks for vulnerabilities and agentic risk

Overview

This skill is a local task-routing memory tool, but it automatically keeps and updates user task metadata without clear opt-in, redaction, retention, or deletion controls.

Review this before installing if you do not want task details, parameters, or scenarios saved across sessions. Use it only if local skill-routing memory is desired, avoid passing secrets or sensitive free-form text into the memory fields, and make sure you have a way to inspect and clear the memory JSON.

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
  • Rogue AgentSelf-Modification, Session Persistence
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (8)

Self-Modification

High
Category
Rogue Agent
Confidence
85% confidence
Finding

Skill modifies its own code, configuration, or behavior at runtime. Self-modification enables an agent to escalate privileges, disable safety constraints, or install persistent backdoors.

Content

Scanner excerpt · SKILL.md (reported line 100)May include surrounding context.

Update Script / 更新脚本

Use the update script to record or update skill memory:

bash
python scripts/update_memory.py --task "任务类型" --skill "技能名称" --params '{"key":"value"}' --scenario "场景描述"

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill explicitly advertises automatic recording and updating of task-to-skill mappings, but it does not disclose to users that task metadata, parameters, and scenarios will be persisted to disk. This creates a privacy and data-retention risk because user-provided content may be stored indefinitely in a workspace file without informed consent or minimization.

Content

No source excerpt is available for this finding.

Ssd 3

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The auto-record design stores task_type, parameters, and scenario directly into persistent memory with no filtering or classification of sensitive input. In practice, user-supplied parameters and scenario text can include secrets, personal data, internal URLs, or other confidential context, making the memory file a durable leakage point.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
86% confidence
Finding

The TTS presets are all fixed to zh-CN voices, which imposes a specific language/locale configuration. The file does not indicate that this is optional, user-selected, or required for a region-specific purpose, so it conflicts with the language/locale policy criterion.

Content

No source excerpt is available for this finding.

Ssd 3

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The workflow instructs the agent to always retain new usage and parameter differences, which causes silent accumulation of potentially sensitive user inputs across sessions. This makes the skill more dangerous because it is not a one-off log entry; it is a long-lived memory mechanism designed to grow automatically.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The workflow's 'always record' and 'always update' directives mandate persistent storage of user task details and parameters without any warning or consent mechanism. Because these fields can contain sensitive requests, tokens, URLs, or personal information, the unconditional retention policy increases privacy exposure over time.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The TTS entry specifies a fixed Chinese voice locale (zh-CN-XiaoyiNeural) and describes it as a Chinese female voice, with no indication that users can choose another language or must opt in to this locale restriction. This is a natural-language policy concern because the configuration appears to enforce a specific language/locale rather than offering choice or documenting a justified region-specific constraint.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
93% confidence
Finding

This code file contains natural-language strings that present the script name, usage examples, and argument examples only in Chinese for key user inputs such as task, skill, and scenario. The file does not offer a language choice or explain a justified locale restriction, which can violate a language/locale policy requiring user opt-in.

Content

No source excerpt is available for this finding.