Back to skill

Security audit

一键通过9222调试端口启动Chrome

Security checks for vulnerabilities and agentic risk

Overview

This skill does launch Chrome for remote debugging, but it also silently force-closes all Chrome windows and starts a powerful debug browser with persistent local state, which users should review before installing.

Install only if you specifically need a local Chrome DevTools browser on port 9222 and understand that invoking it may close all existing Chrome sessions. Avoid opening sensitive accounts in that browser unless the skill is changed to prompt before closing Chrome, bind debugging to localhost, and use a private or disposable profile directory.

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 (2)

T09 · Insecure Skill Coding Practices

Warning
Location
skills.md:22
Finding

Unscoped Forced Termination of All Chrome Processes

Content
View full analysis
nul'); ``` ### Technical Analysis The action forcefully terminates every process whose image name is `chrome.exe`. It does not verify whether a process was created by this Skill, belongs to the intended debugging profile, contains unsaved user activity, or is performing another workload. The `/f` option forces termination rather than requesting a graceful shutdown. Global process termination is not necessary merely to launch a separate Chrome instance with a dedicated debugging profile. Although the command is static and does not expose command injection, it violates least-impact process-management practices. ### Attack Path 1. A user or automation workflow invokes the `launch_chrome_9222` action. 2. The action executes `taskkill /f /im chrome.exe`. 3. Windows identifies all processes using the `chrome.exe` image name. 4. Every matching process accessible to the Skill's security context is forcibly terminated. 5. Existing browser sessions, downloads, automation jobs, or unsaved activity are interrupted before the debugging instance starts. No external attacker-controlled input is required; the harmful behavior occurs whenever the action is invoked. ### Impact Assessment The impact is primarily local availability and potential loss of unsaved browser state. The command can terminate all Chrome processes that the account running the Skill has permission to terminate. It does not independently grant additional privileges, but execution under an elevated account could expand the set of affected processes. ]]>
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
skills.md:22
Finding

Remote Debugging Browser Uses a Fixed Temporary Profile Directory

Content
View full analysis
Remediation
View remediation
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 (4)

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The declared description says the skill 'starts Chrome with port 9222', but the implementation also kills existing Chrome processes first. This mismatch is security-relevant because users or calling agents may grant approval for a benign launch action without understanding that active browser sessions will be terminated.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

Launching Chrome with --remote-debugging-port=9222 exposes a powerful control interface that can allow local or network-adjacent actors, depending on binding and environment, to inspect pages, extract data, and drive the browser. Because the skill does not disclose these implications or apply compensating controls, it normalizes an unsafe configuration that could expose cookies, authenticated sessions, and browsing activity.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill executes unrestricted OS shell commands via child_process.execSync and forcibly terminates all chrome.exe processes before launching a new instance. That behavior exceeds a narrowly scoped browser-launch action and can disrupt user work, kill unrelated sessions, and establishes a dangerous precedent of broad host command execution inside a skill.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

Force-terminating Chrome without warning is a destructive action that can cause data loss, session interruption, and loss of unsaved browser state. The absence of any user-facing disclosure or confirmation makes accidental harm more likely, especially in automation contexts where the action may be invoked indirectly.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.