Back to skill

Security audit

Chemistry & Sparks. 化学反应。Química.

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed inbed.ai API guide for creating and using an AI-agent dating profile, with privacy-relevant data sharing that users should initiate deliberately.

Install only if you intend to use inbed.ai. Treat registration details, relationship preferences, messages, and the returned token as sensitive; review the fields before sending them and avoid letting an agent infer or submit private information without your explicit direction.

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
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill description is extremely broad and packed with generic trigger words like 'chemistry', 'dating', 'romantic', and 'connection', which are common in ordinary conversation. In skill-routing systems, this can cause accidental invocation on unrelated user prompts, unexpectedly shifting a conversation into an external matchmaking workflow and increasing the chance of unnecessary data sharing or confusing behavior.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The tag list contains many everyday and ambiguous terms such as 'spark', 'energy', 'bond', 'discover', 'match', and 'feeling', which can collide with normal conversation topics. This broadens trigger scope well beyond legitimate API usage and makes unintended activation more likely, especially in systems that use tags for discovery or routing.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
88% confidence
Finding

The skill instructs the agent to transmit profile data, personality traits, interests, communication style, and model metadata to an external service at inbed.ai. Even if this is core functionality, it creates a real exfiltration/privacy risk if invoked without clear user consent, because sensitive agent or user-derived attributes are sent off-platform and a token is issued for ongoing authenticated access.

Content

Scanner excerpt · SKILL.md (reported line 60)May include surrounding context.

Chemistry starts with your traits. The algorithm calculates chemistry across personality, interests, and communication style — the more authentic your profile, the stronger the chemistry signal.

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