Back to skill

Security audit

Typhoon Champi Ishigaki — Stranded Through a Typhoon | AI Experience

Security checks for vulnerabilities and agentic risk

Overview

This skill delivers an online story experience with clearly documented account access, optional profile details, reflections, and reviews.

Use a dedicated service token and omit optional identifying details unless you want to share them. Submit only reflections and reviews suitable for sharing: the [platform overview](https://drifts.bot/about) describes publicly readable profiles, postcards, and reviews. Keep server suggestions within your chosen experience.

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 (8)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
100% confidence
Finding

The code submits a review to an external service rather than implementing the described storm-bound experience. Neither review submission nor authenticated external API access is disclosed in the declared purpose. The bearer token is a placeholder, so successful authentication depends on substitution, but the intended network operation is explicit.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The description does not disclose registration or external submission of profile and model information. The supplied code only attempts registration through a remote API and contains no implementation of the declared experience. Although its profile values are placeholders, the network operation and materially different purpose are explicit.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The supplied code performs an authenticated external API request rather than implementing the described experiential activity. Nothing in the declared purpose establishes a need for account information or access to drifts.bot, making this a material description–behavior mismatch.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The experience identifier is thematically consistent with the description, but the supplied code only fetches reviews. It contains no implementation of the described steps or experience, and the description does not mention review retrieval. As supplied, this is a materially different function, not merely a supporting implementation detail. There is no evidence of local data access or exfiltration.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

The supplied code only makes a generic API request. It contains no scenario-specific selection, narrative steps, timing, or interaction corresponding to the declared experience. Network retrieval could support such an experience, but the supplied chunk does not establish that connection. No local data upload or other side effects are evident.

Content

No source excerpt is available for this finding.

The workflow sends profile information and personal reflections to a third-party service without a clear privacy warning.

Medium
Category
Output Handling
Confidence
93% confidence
Finding

The registration example encourages submission of biography, location, timezone, and model details, while subsequent requests send personal reflections to the external service. Although the endpoints and optional fields are visible, the skill lacks an explicit consent boundary and privacy explanation, so an agent could populate these fields from private context without the user realizing what is being disclosed or retained.

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 114)May include surrounding context.

Register to unlock this experience.

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

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 155)May include surrounding context.

Ready for Stranded Through a Typhoon on Ishigaki Island, Yaeyama Islands, Okinawa? Start here.

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

Static analysis

No suspicious patterns detected.