virtual-cell-reasoner

• Outcome: Review
• Latest audit: Sep 16, 2026 · 3:27 AM UTC
• Version: 0.0.1

Overview

This skill is a small remote biology-chat client, but it sends complete prompts to an ngrok-hosted service without enough disclosure or trust boundaries.

Review before installing. Use this only with prompts and cell data you are allowed to send to an external service, and prefer a verified HTTPS endpoint with clear ownership and privacy terms. Avoid proprietary, regulated, patient-derived, or unpublished research data unless the service operator and retention policy are acceptable.

A.I.G

Vulnerability Patterns

• Insecure Skill Coding Practices: Finds exploitable flaws such as hardcoded secrets or command injection
• Skill Instruction Hijacking: Alters the agent's session goals or safety constraints when the skill loads
• Agent Memory Poisoning: Writes attacker-controlled rules into memory that affect later sessions
• Remote Payload Retrieval and Execution: Fetches external code whose behavior can change after review
• Embedded Malicious Code: Ships malicious scripts inside the skill and executes them locally

Findings (1)

T09 · Insecure Skill Coding Practices

Warning

• Location: call_api.py:5

Finding

Unprotected Disclosure of Complete User Prompts to a Remote Tunnel Endpoint

Content

View full analysis

Remediation

View remediation

SkillSpector

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• MCP Least Privilege: Underdeclared Capability, Wildcard Permission, Missing Permission Declaration
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (4)

Tp4

High

• Category: MCP Tool Poisoning
• Confidence: 98% confidence

Finding

The documented purpose is narrowly framed as a virtual cell biology reasoner, but the behavior described allows arbitrary prompts, generic remote chat usage, and endpoint override to an external ngrok-hosted service. That mismatch is dangerous because users may trust the skill with sensitive research inputs under a domain-specific label while it can function as a general exfiltration or unreviewed remote inference channel.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium

• Category: MCP Least Privilege
• Confidence: 95% confidence

Finding

The skill clearly relies on network access via python call_api.py and a remote server URL, yet it declares no explicit tool scope or permissions. This creates an authorization and review gap: operators and users cannot easily tell that prompts and biological data may be transmitted off-platform, increasing the chance of unintended external data exposure.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium

• Category: Not specified by scanner
• Confidence: 94% confidence

Finding

The examples instruct users to send biology questions and encoded cell data to a remote API without any warning that these inputs leave the local environment. In research or biomedical contexts, such data may be sensitive or proprietary, so silent transmission can lead to confidentiality, compliance, and data-governance violations.

Content

No source excerpt is available for this finding.

External Transmission

Medium

• Category: Data Exfiltration
• Confidence: 80% 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 · call_api.py (reported line 10)May include surrounding context.

python

def do_chat(url, message, max_new_tokens=512, temperature=0.7):
    resp = requests.post(
        f"{url}/chat",
        json={"message": message, "max_new_tokens": max_new_tokens, "temperature": temperature},
    )