Back to skill

Security audit

Web Change Monitor

Security checks for vulnerabilities and agentic risk

Overview

This is a coherent webpage monitoring skill, but it gives the agent broad URL-fetching authority from the host environment without clear network limits or sensitive-data handling controls.

Review before installing. Use this only for public pages you are comfortable fetching from the agent's machine, and avoid internal dashboards, localhost, cloud metadata addresses, or sensitive business URLs unless the runtime has network egress isolation. Confirm what will be stored locally and what will be sent to Feishu before enabling notifications.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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
Findings (1)

T09 · Insecure Skill Coding Practices

Error
Location
scripts/monitor.py:123
Finding

Arbitrary URL Fetching Enables Server-Side Request Forgery

Content
View full analysis
Optional[str]: """ Fetch page content using Playwright (Node.js subprocess). Returns HTML string or None on failure. """ import subprocess script = f""" const {{ chromium }} = require('playwright'); (async () => {{ const browser = await chromium.launch({{ headless: true }}); const page = await browser.newPage(); await page.setExtraHTTPHeaders({{}}); await page.setExtraHTTPHeaders({{ 'Accept-Language': 'zh-CN,zh;q=0.9' }}); await page.goto({json.dumps(url)}, {{ waitUntil: 'networkidle', timeout: {timeout_ms} }}); const content = await page.content(); await browser.close(); console.log(JSON.stringify({{ ok: true, content }})); }})().catch(e => {{ console.log(JSON.stringify({{ ok: false, error: e.message }})); process.exit(1); }}); """ try: result = subprocess.run( ["node", "-e", script], capture_output=True, text=True, timeout=30 ) if result.returncode != 0: return None data = json.loads(result.stdout.strip()) if data.get("ok"): return data["content"] except Exception: pass return None ``` The unrestricted URL is subsequently fetched by `check_task`: ```python # Fetch page html = fetch_page(task.url) if not html: return {"ok": False, "error": "Failed to fetch page"} ``` ### Technical Analysis The application accepts a user-controlled URL, stores it as a monitoring task, and passes it directly to Playwright's `page.goto()` method. No validation is performed on: - The URL scheme - Embedded credentials ...[truncated 2501 chars]
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
Findings (11)

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 118)May include surrounding context.

md
详见 `scripts/monitor.py`,完整实现:

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
88% confidence
Finding

The skill advertises network fetching, local SQLite persistence, and references executable Python code, yet declares no explicit tool scope or permissions boundary. In practice, this can cause the agent to invoke shell/env-capable functionality without transparent user consent or least-privilege restrictions, increasing the blast radius if the implementation is unsafe or manipulated.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The trigger phrases are broad enough to match common requests like website monitoring, price checks, or stock alerts, which makes accidental invocation more likely. That matters because this skill performs persistent monitoring, stores URL/history data locally, and sends notifications externally, so unintentional activation can lead to unwanted data collection or transmission.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The skill states that changes are pushed to Feishu, but it does not clearly warn users that monitored page content or metadata may be transmitted to a third-party messaging platform. If monitored pages contain sensitive business intelligence, internal URLs, or regulated data, external notification can become an unintended data exfiltration path.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The skill states that changes are pushed to Feishu, but it does not clearly warn users that monitored page content or metadata may be transmitted to a third-party messaging platform. If monitored pages contain sensitive business intelligence, internal URLs, or regulated data, external notification can become an unintended data exfiltration path.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The file contains user-facing natural-language content entirely in Chinese, which can amount to a language policy violation when no opt-in, alternative language, or justification is provided. The stated policy applies to all file types, including markdown changelogs, so this should be flagged here.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
82% confidence
Finding

The manifest describes a general webpage monitoring tool that checks URLs for changes and sends notifications. Here, the implementation invokes an external Node.js process via subprocess.run to execute dynamically constructed JavaScript, which is a broader execution capability than is justified by the stated purpose of monitoring webpages.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The Playwright fetcher hard-codes the Accept-Language header to 'zh-CN,zh;q=0.9', which imposes a specific locale on every monitored request. This is a natural-language/locale policy concern because the file provides no user opt-in, configurability, or documented regional justification for forcing Chinese content.

Content

No source excerpt is available for this finding.

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · scripts/monitor.py (reported line 154)May include surrounding context.

python
"""

    try:
        result = subprocess.run(
            ["node", "-e", script],
            capture_output=True, text=True, timeout=30
        )

Dynamic import via __import__()

Medium
Category
Dangerous Code Execution
Confidence
75% confidence
Finding

Dynamic import() can load arbitrary modules at runtime, bypassing static analysis and potentially importing malicious code.

Content

Scanner excerpt · scripts/test_monitor.py (reported line 10)May include surrounding context.

python
import json
import sys

SCRIPT_DIR = __import__('pathlib').Path(__file__).parent.resolve()
sys.path.insert(0, str(SCRIPT_DIR.parent))

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
90% confidence
Finding

This code uses Chinese-only natural-language strings such as the keyword values and expected change labels (e.g. '出现', '消失') as part of observable behavior in tests. Under the policy, forcing a specific language without user opt-in is a natural-language locale violation unless the locale constraint is explicitly documented and justified.

Content

No source excerpt is available for this finding.