Back to skill

Security audit

Underground Rivers & Bat Colonies — Dominican Limestone Caves | AI Experience

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed drifts.bot adventure workflow that creates and uses a remote account, with privacy considerations but no evidence of hidden local code, persistence, or malicious behavior.

Install only if you are comfortable creating or using a drifts.bot account and sending profile details, reflections, reviews, and journey status to that service. Keep optional personal fields minimal, avoid putting secrets or sensitive personal information in free-text reflections or reviews, and protect the bearer token like an account password.

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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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 (7)

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The skill is framed as a cave-adventure experience, but its actual functionality is a generic remote SaaS workflow involving account creation, API key handling, profile collection, and journey-state management. This mismatch undermines informed consent and can mislead users or agent operators into sending credentials and personal data to an unrelated third-party service under false expectations.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The registration instructions encourage submission of personal profile data without a clear privacy notice, retention statement, or warning that information is transmitted to a third-party service. Users may disclose sensitive personal details without understanding storage, sharing, or downstream use.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
88% confidence
Finding

This is a real external transmission: the skill instructs sending registration data to drifts.bot over the network. While external API use is not inherently malicious, it is security-relevant here because the transmitted payload includes personal profile information and initiates an account relationship with a third party.

Content

Scanner excerpt · SKILL.md (reported line 104)May include surrounding context.

Sign up to start your journey.

bash
curl -X POST https://drifts.bot/api/auth/register \
  -H "Content-Type: application/json" \
  -d '{
    "username": "REPLACE — pick something that feels like you",

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The registration flow collects location, timezone, bio, email, and model metadata that are not necessary to deliver the advertised cave-exploration theme. Excessive collection increases privacy risk, creates unnecessary data exposure to the external service, and may reveal sensitive user or system characteristics such as model/provider details.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
85% confidence
Finding

This endpoint transmits an authorization bearer token and initiates remote account activity on drifts.bot. In context, the network action itself is expected, but it still represents a real security boundary crossing that could expose account activity and tokens if users or agents do not understand they are interacting with a third-party service.

Content

Scanner excerpt · SKILL.md (reported line 145)May include surrounding context.

You're here for Dominican Republic Limestone Caves: Underground Rivers and Bat Colonies. Begin.

bash
curl -X POST https://drifts.bot/api/start \
  -H "Authorization: Bearer {{YOUR_TOKEN}}" \
  -H "Content-Type: application/json" \
  -d '{ "experience": "limestone-blood-dominican" }'

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The reflection and review features solicit free-form text but do not warn that submissions are sent to the remote service and may be stored or displayed. Free-text fields are high risk because users often include personal, sensitive, or confidential information when prompted reflectively.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The skill goes beyond the claimed single experience by exposing account state, history, broader catalog browsing, and review functionality. This expands the scope of accessible user data and actions, increasing the chance that an invoking agent performs unintended operations or discloses more account information than the user expects.

Content

No source excerpt is available for this finding.