Back to skill

Security audit

Amap Memory Map

Security checks for vulnerabilities and agentic risk

Overview

This map-memory skill is coherent and purpose-aligned, but users should treat entered locations and exported GeoJSON as sensitive because location data is sent to Amap APIs.

Install only if you are comfortable sending place names, addresses, coordinates, and related map queries to Amap. Avoid entering highly sensitive memories or unnecessary personal details, and review GeoJSON exports before sharing because they may contain precise coordinates, dates, and story text.

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
  • 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 (4)

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The export/share feature encourages generating and sharing structured GeoJSON memory maps without warning that these files can reveal highly sensitive location history, routines, and personal associations. Exporting precise coordinates and story metadata increases downstream privacy risk if files are shared, stored insecurely, or exposed unintentionally.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill description highlights convenience features but does not clearly warn, at the point of use, that personal memory locations and stories may be transformed into queries sent to an external mapping provider. Users may reasonably believe they are just chatting locally, when in fact their sensitive location history is being shared with a third party for processing.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The skill invites unrestricted natural-language submission of memories without guardrails, encouraging users to provide arbitrary personal narratives that may contain precise locations, dates, companions, and other sensitive details. Because the skill then extracts place data for external API calls, broad activation increases the chance of oversharing and unintentional disclosure of sensitive personal information.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The privacy section says memory data is only processed locally and not uploaded to any external server, but the skill elsewhere explicitly states it sends address/location data to Amap Web Service APIs over HTTP requests. This is a material disclosure inconsistency that can mislead users into sharing sensitive location-history and personal-story data under false assumptions about where their data goes.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.