ClawHub
Back to skill

Security audit

EverMemory

Security checks for vulnerabilities and agentic risk

Overview

EverMemory is a disclosed persistent-memory skill with no bundled executable code, and its sensitive capabilities are aligned with its stated purpose.

Install this only if you want durable memory behavior. Expect it to store and recall facts, preferences, project context, profiles, and behavior rules; review exports/imports/restores carefully and avoid saving secrets unless you intentionally want them persisted.

A.I.G

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

SkillSpector

Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • 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 (3)

Vague Triggers

Medium
Confidence
93% confidence
Finding
The activation condition is very broad and covers many common memory-related topics, which can cause the skill to trigger in situations where the user did not clearly intend persistent memory operations. In this skill’s context, unintended invocation is more concerning because the toolset includes durable storage, profile building, and export/import actions, so accidental activation can lead to unnecessary collection or exposure of sensitive information.

Natural-Language Policy Violations

Medium
Confidence
95% confidence
Finding
The `when` field is written only in Chinese, which effectively constrains invocation policy to a specific language without any stated opt-in or justification. This can violate language/locale policy expectations for multilingual environments, especially since the rest of the document is largely in English.

Missing User Warnings

Medium
Confidence
89% confidence
Finding
The skill describes persistent memory, profile generation, export/import, and governed self-improvement without presenting an upfront warning that these operations may involve sensitive personal or project data. In this context, missing disclosure increases the risk that users will unknowingly authorize storage, profiling, backup, or migration of confidential information.

Static analysis

No suspicious patterns detected.