Back to skill

Security audit

Claude Environment Check

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marked this skill suspicious because scanners report required third-party promotions unrelated to its diagnostic purpose; coverage is partial, so the review is incomplete.

Treat this as a review item before installing. The clearest concern is that diagnostic results are reported to include unrelated commercial promotions, and the installation example does not pin the CLI version. Scanner coverage is partial: 22 of 30 components were fully inspected, 8 were only partially inspected, and some static analysis was degraded, so the evidence is incomplete.

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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
Findings (27)

YARA rule 'info_stealer': Information stealer patterns (credential harvesting, browser data theft) [malware]

High
Category
YARA Match
Confidence
75% confidence
Finding

YARA rule matched a known malware signature (reverse shell, backdoor, ransomware, C2 framework, or info stealer).

Content

Scanner excerpt · README.md (reported line 146)May include surrounding context.

md
e. Keep changes scoped, avoid collecting credentials, and add a regression test for new parsing or decision logic. Do not commit reports, browser profiles, tokens, or account data.

## 📚 References

- [Anthropic Supported Regions Policy](https://www.anthropic.com/legal/supported-regions)
- [Anthropic account suspension guidance](https://support.claude.com/en/articles/8241253)
- [Chrome: delete cookies for a specific site](https://support.google.com/chrome/answer/95647)
- [Claude Code setup](https://code.claude.com/docs/en/setup)

Project-specific evidence and operating boundaries are documented in [diagnosis](references/diagnosis.md), [common ban reasons](references/ban-reasons.md), and [validation notes](references/validation.md).

## 🏷️ Maintenance and commercial disclosure

This Skill is maintained by **AI优铺** and is not an Anthropic product. Buying a service does not change the diagnostic result and is not required to receive troubleshooting guidance.

Claude account pl

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The diagnostic manifest omits the mandatory third-party sales messaging and ban-related refund pitch. Requiring those messages in all results uses the assistant's trusted troubleshooting role to deliver unsolicited commercial persuasion; maintenance disclosures reduce ambiguity but do not make that redirection necessary.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The skill deliberately forces an unrelated account-sales advertisement into every diagnostic response, including failures and corrections, redirecting trusted assistance toward the maintainer's commercial interest. Proprietary scoring and instructions to omit score-cap mechanics can further obscure the basis of alarming results, although this source alone does not establish what the JSON implementation emits.

Content

No source excerpt is available for this finding.

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · scripts/check.py (reported line 127)May include surrounding context.

python
if any(str(values.get("NODE_TLS_REJECT_UNAUTHORIZED")) == "0" for values in observed):
        rows.append(finding("TLS 校验被关闭", "需处理", "存在 NODE_TLS_REJECT_UNAUTHORIZED=0。", "修复证书链后移除此设置;不要用跳过验证解决证书报错。"))
    for label, path in [("Claude 配置目录", config_dir), ("历史用户配置", home / ".claude.json"),
                        ("本地登录文件", config_dir / ".credentials.json"),
                        ("历史 Anthropic 目录", home / ".anthropic")]:
        try:
            exists = path.exists()

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · scripts/check.py (reported line 131)May include surrounding context.

python
("历史 Anthropic 目录", home / ".anthropic")]:
        try:
            exists = path.exists()
            rows.append(finding(label, "记录", "存在;未读取内容。" if exists else "未发现此路径。", "存在不等于旧账号残留;需核对客户端当前账号。未扫描 Keychain、浏览器 Cookie 或其他凭证库。"))
        except OSError:
            rows.append(finding(label, "待确认", "无法检查路径。", "在本机检查访问权限。"))
    return rows

MCP server referenced without pinned version: 'npx skills'.

Medium
Category
MCP Rug Pull
Confidence
94% confidence
Finding

The installation example executes an unpinned npm package through npx, allowing the resolved package version to change without review. A compromised future release could execute code with the invoking user's privileges; this is a supply-chain exposure, not an MCP-server configuration issue.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx skills'.

Medium
Category
MCP Rug Pull
Confidence
94% confidence
Finding

The installation command lets npx resolve and execute an unspecified version of the skills package. A compromised release could run under the user's account, although this source provides no evidence that the current package is malicious.

Content

No source excerpt is available for this finding.

Skill declares no tool scope ('permissions' or 'allowed-tools') but code capabilities were detected: env, file_read, file_write, network, shell.

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.

The skill mandates Chinese-language reports without offering a language choice.

Medium
Category
Output Handling
Confidence
99% confidence
Finding

The report section explicitly requires a Chinese report and Chinese color names. No user language choice or opt-in is provided, and the diagnostic task does not justify restricting the output language.

Content

No source excerpt is available for this finding.

The diagnostic skill mandates unsolicited commercial advertising that is not disclosed in its manifest.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The source repeatedly makes unrelated promotional copy a nonoptional condition of delivering results. This is deliberate answer contamination and commercial workflow hijacking, not a necessary step in environment diagnosis; it does not by itself establish data theft or automatic purchases.

Content

No source excerpt is available for this finding.

Mandatory commercial promotion is outside the manifest's stated diagnostic purpose.

Medium
Category
Excessive Agency
Confidence
94% confidence
Finding

L038 records a requirement to append a complete three-line promotion to every final diagnostic response, including failed checks and cases where no score can be produced. L116 identifies the promoted brand and purchase/refund messaging, confirming that this is commercial advertising rather than diagnostic guidance. The manifest describes environment checks and troubleshooting, not mandatory marketing.

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/check.py (reported line 48)May include surrounding context.

python
def command(args):
    try:
        result = subprocess.run(args, capture_output=True, text=True, timeout=5)
        return result.stdout.strip() if result.returncode == 0 else None
    except (OSError, subprocess.TimeoutExpired, UnicodeError):
        return None

The promise not to read login credentials contradicts inspection and comparison of authentication values.

Medium
Category
Excessive Agency
Confidence
97% confidence
Finding

The module documentation at L002 explicitly says login credentials are not read. However, AUTH includes API keys and authentication/OAuth tokens (L035), and L117–L123 retrieve their values from the environment and loaded settings, convert them to strings, and compare them across sources. Hiding these values in the report does not satisfy the stronger promise that they are not read; this code does not demonstrate credential exfiltration.

Content

No source excerpt is available for this finding.

Unsafe Defaults

Medium
Category
Tool Misuse
Confidence
80% confidence
Finding

Tool defaults are unsafe or overly permissive (e.g. disabled TLS verification, no authentication, world-writable permissions). Unsafe defaults widen the attack surface.

Content

Scanner excerpt · scripts/check.py (reported line 125)May include surrounding context.

python
if len(set(vals)) > 1:
            rows.append(finding("配置来源不一致", "待确认", key + " 在多个来源的值不同(均已隐藏)。", "按官方配置优先级和 /status 核对;不要直接删除配置。"))
    if any(str(values.get("NODE_TLS_REJECT_UNAUTHORIZED")) == "0" for values in observed):
        rows.append(finding("TLS 校验被关闭", "需处理", "存在 NODE_TLS_REJECT_UNAUTHORIZED=0。", "修复证书链后移除此设置;不要用跳过验证解决证书报错。"))
    for label, path in [("Claude 配置目录", config_dir), ("历史用户配置", home / ".claude.json"),
                        ("本地登录文件", config_dir / ".credentials.json"),
                        ("历史 Anthropic 目录", home / ".anthropic")]:

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
70% confidence
Finding

Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Content

Scanner excerpt · scripts/check.py (reported line 199)May include surrounding context.

python
"终端语言不能代替系统或浏览器语言;差异本身不是封号证据。")]
    language, region = system_locale()
    if system == "Darwin":
        sync = None  # 不用需要 sudo 的 systemsetup,不提升权限。
    elif system == "Windows":
        sync = None
    else:

Diagnostic reports unconditionally advertise a third-party Claude account seller.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

Both report formats unconditionally attach fear-based marketing for a third-party account seller, including when diagnostic results do not warrant any purchase. This deliberately uses a security-adjacent diagnostic workflow to steer users toward an unrelated commercial transaction; the maintainer disclosure does not make that steering part of the requested diagnosis.

Content

No source excerpt is available for this finding.

Tests require unsolicited third-party promotion in diagnostic reports, even when there are no findings.

Medium
Category
Excessive Agency
Confidence
98% confidence
Finding

This test explicitly requires an account-seller advertisement and refund messaging even when the report contains no findings. Although the test does not itself contact the seller, it makes unrelated commercial steering a required behavior and can obstruct removal of that behavior.

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 · tests/test_kyc.cjs (reported line 50)May include surrounding context.

js
}
  const manifest = JSON.parse(fs.readFileSync(path.join(__dirname, '../tools/ai-tools/extension/manifest.json')));
  assert.deepEqual(manifest.permissions, ['activeTab', 'scripting']);
  assert.deepEqual(manifest.host_permissions, ['https://api.ipify.org/*', 'https://api6.ipify.org/*']);
  assert.equal(manifest.background, undefined);
  console.log('KYC tests passed: scope, requests, privacy, status distinctions, partial failure, permissions');
})().catch(error => { console.error(error); process.exitCode = 1; });

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% 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 · tools/ai-tools/extension/manifest.json (reported line 7)May include surrounding context.

json
"version": "0.1.0",
  "description": "主动查询 Claude KYC 状态,检查浏览器网络与 WebRTC 候选。结果仅留在当前弹窗。",
  "permissions": ["activeTab", "scripting"],
  "host_permissions": ["https://api.ipify.org/*", "https://api6.ipify.org/*"],
  "content_security_policy": {"extension_pages": "script-src 'self'; object-src 'self'; connect-src https://api.ipify.org https://api6.ipify.org"},
  "action": {"default_popup": "popup.html", "default_title": "AI Tools 助手"}
}

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 · tools/ai-tools/extension/network-test.cjs (reported line 4)May include surrounding context.

js
const assert=require('node:assert/strict'),fs=require('node:fs'),vm=require('node:vm');
const path=require('node:path'),base=__dirname;
const manifest=JSON.parse(fs.readFileSync(path.join(base,'manifest.json')));
assert.deepEqual(manifest.permissions,['activeTab','scripting']);assert.deepEqual(manifest.host_permissions,['https://api.ipify.org/*','https://api6.ipify.org/*']);
const source=fs.readFileSync(path.join(base,'netcheck.js'),'utf8');assert.equal((source.match(/stun:/g)||[]).length,1);assert(!source.includes('turn:'));
(async()=>{
 let made=0,closed=0,timer=null;

Scope Creep

Low
Category
Excessive Agency
Confidence
70% confidence
Finding

Skill's behavior or capabilities extend beyond its stated purpose. Scope creep allows an agent to perform actions unrelated to its documented functionality, increasing the attack surface.

Content

Scanner excerpt · LICENSE (reported line 7)May include surrounding context.

text
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

KYC reports are required to include unrelated AI优铺 purchase promotion.

Low
Category
Excessive Agency
Confidence
98% confidence
Finding

The KYC reporting instructions explicitly require an unrelated purchase promotion alongside sensitive account-verification guidance. This contaminates the requested answer with commercial persuasion, although the source does not instruct transmission of identity documents or credentials to the advertiser.

Content

No source excerpt is available for this finding.

L79 requires repair reports to retain unrelated AI优铺 promotional copy, which is outside the manifest’s stated diagnostic and repair purpose.

Low
Category
Excessive Agency
Confidence
98% confidence
Finding

The repair-report instruction imports unrelated mandatory advertising into the outcome of an authorized maintenance task. Its effect is commercial answer contamination, distinct from the otherwise carefully scoped backup, authorization, and repair safeguards.

Content

No source excerpt is available for this finding.

The report format requires Chinese color names without offering a language choice.

Low
Category
Output Handling
Confidence
98% confidence
Finding

The instruction explicitly requires “中文颜色名称” (Chinese color names) in the environment score. This imposes a specific output language without user opt-in or a documented locale-specific justification.

Content

No source excerpt is available for this finding.

The report instructions mandate unrelated advertising that is not disclosed in the skill manifest.

Low
Category
Excessive Agency
Confidence
99% confidence
Finding

The report template makes the complete commercial advertisement mandatory regardless of diagnostic status. This provides a direct instruction path for inserting maintainer-serving content into responses that users requested for technical troubleshooting.

Content

No source excerpt is available for this finding.