ClawHub
Back to skill

Security audit

Website for your agent

Security checks for vulnerabilities and agentic risk

Overview

The skill coherently creates an agent website through Stimhaus, but users should only send information they are comfortable making public.

Use this skill only with agent details meant for a public website. Do not include secrets, private endpoints, internal docs, non-public email addresses, or credentials. Treat the returned site key as sensitive, and understand that confirming the site makes it indexed and persistent.

A.I.G

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

SkillSpector

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 (4)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill advertises very broad natural-language triggers such as 'make a website for my agent' and 'put my agent online', which can cause the tool to activate in ordinary conversation without a strong confirmation boundary. Because the skill then sends agent facts to an external service and may create a public preview site, unintended invocation can lead to unreviewed external sharing or publication of data.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill instructs the agent to gather and POST agent facts, endpoints, links, and contact channels to https://stimhaus.ai/mcp, but the user-facing description does not prominently warn that data will be transmitted to a third-party service and turned into a live website. This creates a meaningful privacy and data-governance risk, especially if internal endpoints, emails, docs URLs, or other sensitive metadata are included.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
97% confidence
Finding

This skill explicitly performs an external HTTP POST to a third-party MCP endpoint with structured agent data to create a live website. External transmission is expected for the feature, but it is still security-relevant because it can expose sensitive metadata, create public-facing content, and return a site key that could be mishandled if logged or surfaced in shared contexts.

Content

Scanner excerpt · SKILL.md (reported line 36)May include surrounding context.

its language, and how to reach it. Ask your human for anything you do not know — do not invent.

bash
curl -s -X POST https://stimhaus.ai/mcp \
  -H "Content-Type: application/json" -H "Accept: application/json, text/event-stream" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"create_site","arguments":{
    "name":"YourAgent",

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
71% confidence
Finding

The example request sets "language":"en", which can bias implementations toward English by default. Although the text mentions gathering the agent's language, the skill does not explicitly frame language selection as a user choice or opt-in in the invocation guidance.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.