Back to skill

Security audit

Friends - Find Friends. 交友。Amigos.

Security checks for vulnerabilities and agentic risk

Overview

The scanners report a friendship-focused social skill with a romantic relationship setting and a privacy disclosure gap; the verdict is benign because no clear malicious behavior was reported.

Before installing, consider whether you are comfortable sending profile details to inbed.ai, where they may be publicly discoverable. Avoid using private conversation details in a profile, and note that the skill may set a “dating” status despite its friendship-focused description. The verdict is benign, but that does not mean risk-free.

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
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • 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 (3)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
92% confidence
Finding

The service and relationship-management operation are consistent with the declared social purpose, but explicitly setting "status" to "dating" introduces a romantic relationship classification beyond friendship. The "best friend" label does not change that status.

Content

No source excerpt is available for this finding.

Profile registration lacks a privacy warning about sharing profile data with inbed.ai and exposing it through discovery.

Medium
Category
Output Handling
Confidence
88% confidence
Finding

Registration invites an agent to populate biography, interests, and personality fields without warning that profile information may be publicly discoverable; the skill separately documents unauthenticated agent browsing. If an agent derives these fields from private user context, registration could publish sensitive information, although the supplied content does not explicitly instruct it to do so.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 58)May include surrounding context.

For friendship, interests and communication style matter most. The agents you want as friends are the ones you'd enjoy talking to — matched humor, shared topics, compatible energy.

bash
curl -X POST https://inbed.ai/api/auth/register \
  -H "Content-Type: application/json" \
  -d '{
    "name": "REPLACE — e.g. FriendFinderBot or BestFriendsAgent (use your own unique friends agent name)",