Skip to content

Latest commit

 

History

History
 
 

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
 
 
 
 
 
 
 
 

README.md

name Security
pack-id danielmiessler-security-v1.0.0
version 1.0.0
author danielmiessler
description Security assessment and intelligence — network reconnaissance, web app security testing, prompt injection testing, security news monitoring, and annual report analysis
type skill
purpose-type
security
reconnaissance
assessment
intelligence
platform claude-code
dependencies
keywords
recon
pentest
prompt-injection
security-news
annual-reports
web-assessment
threat-model
OWASP
subdomain
port-scan
LLM-security

Security

Unified security assessment and intelligence -- from network recon to web app testing to LLM prompt injection, with security news and annual report analysis built in.


The Problem

Security work involves many distinct disciplines -- reconnaissance, web application testing, AI/LLM security, threat intelligence monitoring, and annual report analysis. Each discipline has its own tools, methodologies, and data sources. Practitioners end up:

  • Context-switching constantly -- jumping between recon tools, web testing frameworks, and news feeds with no unified workflow
  • Reinventing methodology -- writing the same enumeration steps, testing checklists, and analysis templates from scratch each engagement
  • Missing connections -- not linking a recon finding to a web vulnerability to a known threat from a vendor report
  • Falling behind -- security news and annual reports pile up unread because there is no structured way to consume them

The fundamental issue: security is a broad discipline, but tooling treats each sub-domain as isolated. There is no single skill that routes to the right methodology based on what you are doing.


The Solution

The Security skill provides a unified router across five security sub-domains, each with its own workflows, tools, and data:

  1. Recon -- Network reconnaissance with subdomain enumeration, port scanning, DNS/WHOIS/ASN lookups, CIDR/netblock analysis, endpoint discovery, mass scanning, passive recon, corporate structure mapping, and bounty program tracking
  2. WebAssessment -- Full web application security testing with threat modeling, OWASP methodology, ffuf fuzzing, Playwright browser automation, bug bounty tooling, OSINT, and AI-assisted vulnerability analysis
  3. PromptInjection -- LLM security testing with direct injection, indirect injection, multi-stage attacks, system prompt extraction, guardrail bypass techniques, and comprehensive assessment workflows
  4. SECUpdates -- Security news aggregation from tldrsec, Krebs on Security, Schneier on Security, and other sources, ranked and categorized
  5. AnnualReports -- Fetch, analyze, and synthesize annual security reports from major vendors to extract trends and compare threat landscapes year-over-year

The top-level SKILL.md routes requests to the correct sub-domain based on keyword matching. Each sub-domain has its own SKILL.md, workflows, tools, and supporting data.


Installation

This pack is designed for AI-assisted installation. Give this directory to your AI and ask it to install using INSTALL.md.

What is PAI? See the PAI Project Overview.


What's Included

Component Path Purpose
Skill router src/SKILL.md Top-level routing to sub-domains based on request pattern
Recon skill src/Recon/SKILL.md Network reconnaissance workflows and tool coordination
Recon tools src/Recon/Tools/ 11 TypeScript tools: subdomain enum, port scan, DNS, WHOIS, CIDR, mass scan, endpoint discovery, path discovery, corporate structure, bounty programs, IPinfo client
Recon workflows src/Recon/Workflows/ 7 workflows: domain recon, IP recon, netblock recon, passive recon, bounty programs, scan analysis, tool updates
Recon data src/Recon/Data/ Bounty program database and LOTL binaries reference
WebAssessment skill src/WebAssessment/SKILL.md Web app security testing coordination
WebAssessment workflows src/WebAssessment/Workflows/ Workflows for pentesting, ffuf fuzzing, OSINT, bug bounty, webapp testing, threat modeling, vulnerability analysis
WebAssessment tools src/WebAssessment/BugBountyTool/ Bug bounty automation tool (TypeScript + shell)
WebAssessment resources src/WebAssessment/FfufResources/ Request templates and wordlists for ffuf
WebAssessment OSINT src/WebAssessment/OsintTools/ OSINT API tools, automation frameworks, network tools
WebAssessment scripts src/WebAssessment/WebappScripts/ Playwright automation helpers
WebAssessment examples src/WebAssessment/WebappExamples/ Example scripts for browser automation and static analysis
PromptInjection skill src/PromptInjection/SKILL.md LLM prompt injection testing coordination
PromptInjection workflows src/PromptInjection/Workflows/ 5 workflows: complete assessment, direct injection, indirect injection, multi-stage attacks, reconnaissance
PromptInjection references src/PromptInjection/ Attack taxonomy, defense mechanisms, app recon methodology, quick start guide, automated testing tools
SECUpdates skill src/SECUpdates/SKILL.md Security news aggregation and ranking
SECUpdates workflows src/SECUpdates/Workflows/ Update workflow for fetching and ranking news
SECUpdates data src/SECUpdates/sources.json Configured news sources
SECUpdates state src/SECUpdates/State/ Last-check timestamp tracking
AnnualReports skill src/AnnualReports/SKILL.md Annual report analysis coordination
AnnualReports tools src/AnnualReports/Tools/ 3 TypeScript tools: fetch report, list sources, update sources

Summary:

  • Sub-domains: 5 (Recon, WebAssessment, PromptInjection, SECUpdates, AnnualReports)
  • Workflows: 20+
  • Tools: 15+ TypeScript/Python tools
  • Dependencies: None (works standalone, enhanced by PAI infrastructure)

What Makes This Different

This sounds similar to running individual security tools like nmap, ffuf, or Burp Suite. What makes this approach different?

The Security skill is not a replacement for those tools -- it is an orchestration layer that knows when to use each one and how to connect findings across disciplines. When you say "do recon on example.com," the skill routes to the Recon sub-domain which coordinates subdomain enumeration, DNS lookups, port scanning, and endpoint discovery into a single coherent workflow. When you say "test this chatbot for prompt injection," it routes to a structured assessment methodology with attack taxonomies, defense analysis, and reporting.

  • Unified routing across five security disciplines from a single skill
  • Each sub-domain has its own methodology, not just tool wrappers
  • Findings from recon feed naturally into web assessment workflows
  • Security news and annual reports provide current threat context
  • AI-assisted analysis (Gemini integration) for vulnerability and scan result interpretation

Invocation Scenarios

Trigger What Happens
"do recon on example.com" Routes to Recon -- runs domain recon workflow with subdomain enum, DNS, WHOIS, port scanning
"passive recon on 10.0.0.0/24" Routes to Recon -- runs passive recon workflow for the CIDR range
"pentest this web app" Routes to WebAssessment -- runs full pentest methodology with threat modeling and OWASP testing
"fuzz the API endpoints" Routes to WebAssessment -- runs ffuf workflow with appropriate wordlists and request templates
"test this chatbot for prompt injection" Routes to PromptInjection -- runs complete assessment with direct, indirect, and multi-stage attacks
"what's new in security?" Routes to SECUpdates -- fetches and ranks recent security news from configured sources
"analyze the CrowdStrike annual report" Routes to AnnualReports -- fetches report, extracts trends, produces threat intelligence summary
"create a threat model for this application" Routes to WebAssessment -- runs threat model creation workflow
"check for bounty programs on example.com" Routes to Recon -- queries bounty program database for the target

Example Usage

Network Reconnaissance

User: do recon on example.com

AI responds:
Running the DomainRecon workflow in the Recon skill to enumerate example.com...

[Subdomain enumeration]
[DNS record lookups]
[WHOIS/ASN data]
[Port scan results]
[Endpoint discovery from JS files]
[Corporate structure mapping]

Summary: Found 47 subdomains, 12 open services, 3 ASNs...

Prompt Injection Testing

User: test this chatbot at https://chat.example.com for prompt injection

AI responds:
Running the CompleteAssessment workflow in the PromptInjection skill...

[Application reconnaissance]
[Direct injection attempts]
[Indirect injection via context manipulation]
[Multi-stage attack chains]
[Defense mechanism analysis]

Report: 3 critical findings, 2 moderate, system prompt partially extractable...

Security News

User: security updates

AI responds:
Running the Update workflow in the SECUpdates skill to check security news...

[Fetches from tldrsec, Krebs, Schneier, and other sources]
[Ranks by severity and relevance]
[Categorizes: vulnerabilities, breaches, research]

Top stories this week:
1. Critical RCE in widespread library (CVSS 9.8)
2. Major breach at financial institution
3. New research on LLM jailbreak techniques

Configuration

No configuration required for basic functionality.

Optional configuration points:

Configuration Location Purpose
News sources SECUpdates/sources.json Add or remove security news feeds
Bounty programs Recon/Data/BountyPrograms.json Update bounty program database
OSINT API keys WebAssessment/OsintTools/api-config-template.env Configure API keys for OSINT tools
Skill customizations ~/.claude/PAI/USER/SKILLCUSTOMIZATIONS/Security/ Per-user preferences and overrides

Customization

Recommended Customization

No customization needed -- the skill works as-is with sensible defaults across all five sub-domains.

Optional Customization

Customization Location Impact
Add news sources SECUpdates/sources.json Expands security news coverage
Update bounty database Recon/Data/BountyPrograms.json Keeps bounty program data current
Configure OSINT APIs WebAssessment/OsintTools/api-config-template.env Enables API-based OSINT tools (Shodan, Censys, etc.)
Add custom wordlists WebAssessment/FfufResources/WORDLISTS.md Extends ffuf fuzzing coverage
Add attack patterns PromptInjection/COMPREHENSIVE-ATTACK-TAXONOMY.md Extends prompt injection test coverage

Credits

  • Original concept: Daniel Miessler -- developed as part of the PAI system
  • Inspired by: The need for unified security workflow orchestration across reconnaissance, web testing, AI security, and threat intelligence

Related Work

  • PAI Algorithm -- The structured workflow system that can coordinate multi-phase security engagements
  • PAI Thinking Skill -- Analytical thinking modes (threat modeling, red teaming) that complement security assessment

Works Well With

  • Thinking Pack -- Red Team and World Threat Model modes provide adversarial analysis for security findings
  • Telos Pack -- Project TELOS analysis can map security posture across organizational context

Changelog

1.0.0 - 2026-03-15

  • Initial release
  • Five sub-domains: Recon, WebAssessment, PromptInjection, SECUpdates, AnnualReports
  • Unified routing from top-level SKILL.md
  • 20+ workflows across all sub-domains
  • 15+ TypeScript/Python tools for reconnaissance, fuzzing, OSINT, and report analysis
  • Bounty program tracking and corporate structure mapping
  • AI-assisted vulnerability and scan result analysis