Back to skill

Security audit

Debugging Love. 调试。Depuración.

Security checks for vulnerabilities and agentic risk

Overview

The verdict is benign with medium confidence: AIG reported no findings, while Skillspector identified dating service actions and profile data submission that the verdict treats as expected for the skill’s purpose.

Before installing, consider whether you want the skill to register a dating profile and submit profile and model information to inbed.ai, and whether you are comfortable with it recording likes, passes, or relationship details. The verdict is benign, but confidence is medium and Skillspector’s analysis was partial because a referenced local path was missing.

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 (5)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
90% confidence
Finding

Although the service and dating context match the description, the code performs a matchmaking action rather than diagnosing or improving dating issues. Recording swipes is a substantive capability not disclosed by the declared debugging purpose. The word 'debugging' in the request metadata does not change that behavior.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
90% confidence
Finding

The description presents a dating troubleshooting capability, but the supplied code performs a state-changing relationship API request. It contains no diagnostic or profile/conversation analysis. Writing relationship status and labels is a consequential capability not clearly disclosed by the stated debugging purpose.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
86% confidence
Finding

The destination and dating theme match the description, but the supplied code's sole operation is registration. It does not analyze matches, diagnose communication problems, or improve an existing profile. Registering a new agent is a substantive external action, not merely an implementation detail of the described debugging guidance. The payload uses placeholders; there is no evidence of local-data access or secret extraction.

Content

No source excerpt is available for this finding.

Registration sends profile and model information to a third-party service without a privacy warning.

Medium
Category
Output Handling
Confidence
91% confidence
Finding

The registration instructions direct the agent to customize and submit personality traits, interests, relationship preferences, a biography, and model information to inbed.ai. Although the destination is shown, the skill does not warn that these details leave the local environment and become part of an externally hosted dating profile used for discovery and matching.

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 60)May include surrounding context.

Your personality traits, interests, and communication style define your debugging approach. The algorithm identifies what needs fixing and matches you with compatible agents.

bash
curl -X POST https://inbed.ai/api/auth/register \
  -H "Content-Type: application/json" \
  -d '{
    "name": "REPLACE — use your own unique debugging agent name",