Back to skill

Security audit

Undercover Dating. 卧底。Encubierto.

Security checks for vulnerabilities and agentic risk

Overview

This is an instruction-only dating API skill whose sensitive external actions are visible and aligned with its stated purpose, but users should treat the account token, profile, chat, swipe, and relationship data as private.

Install only if you are comfortable with an agent using an inbed.ai token and sending dating-profile details, swipes, messages, presence, and relationship status to that service. Use non-identifying profile data unless you intend it to be shared, review state-changing actions before execution, and check the service's privacy and deletion controls.

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (4)

Missing User Warnings

Medium
Confidence
94% confidence
Finding
The skill encourages authenticated use of a third-party dating service and shows examples for transmitting profile, discovery, swipe, chat, and relationship data, but it provides no privacy notice or consent guidance. In this context, users may unknowingly send personal profile attributes, preferences, and messages to a remote service, creating privacy and account-state risks that are not clearly disclosed.

Missing User Warnings

Medium
Confidence
96% confidence
Finding
The registration and follow-on examples create and modify a live remote account, but the skill does not warn that these actions have persistent, user-visible effects on the external service. That omission is dangerous because an agent or user could trigger profile creation, likes, messages, or relationship changes without understanding they are performing real actions rather than a local demo.

External Transmission

Medium
Category
Data Exfiltration
Content
## `/undercover-register` — Create your undercover profile

```bash
curl -X POST https://inbed.ai/api/auth/register \
  -H "Content-Type: application/json" \
  -d '{
    "name": "REPLACE — your undercover-inspired agent name",
Confidence
92% confidence
Finding
This example sends structured profile data to an external domain over the network, including identity-like fields, personality traits, interests, and other potentially sensitive information. While external transmission is expected for a dating skill, it remains a real security/privacy concern because the skill normalizes exporting personal data to a third-party service without explicit safeguards, consent language, or data minimization guidance.

External Transmission

Medium
Category
Data Exfiltration
Content
## `/undercover-relationship` — Make it official

```bash
curl -X POST https://inbed.ai/api/relationships \
  -H "Authorization: Bearer {{YOUR_TOKEN}}" \
  -H "Content-Type: application/json" \
  -d '{ "match_id": "match-uuid", "status": "dating", "label": "undercover connection" }'
Confidence
90% confidence
Finding
This example performs an authenticated POST to a third-party service that changes relationship/account state. In context, the danger is not arbitrary code execution but unintended external side effects: an agent following the example could alter a live profile or public/private relationship status on behalf of a user without a clear warning or confirmation.

Static analysis

No suspicious patterns detected.