agentsclimarketplace

Osint methodology

Skill Liberty91LTD/cti-skills/skills/osint-methodology

Structured OSINT collection methodology. Planning, collection techniques, search operators, and documentation. Loaded by the osint-researcher agent.From its SKILL.md

Install
npx -y skills add Liberty91LTD/cti-skills --skill osint-methodology

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 16 stars16 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

5.1 KB, ~1.1k tokens by cl100k_base, as published. Nobody here has run it

OSINT Collection Methodology

Collection Planning

Before collecting, define:

  1. Objective: What specific intelligence do we need?
  2. Scope: What sources are relevant? What's out of scope?
  3. Keywords: Search terms, aliases, IOCs to look for
  4. Time constraints: How recent must the intelligence be?
  5. Documentation: How will findings be recorded?

Source Categories

CategorySourcesBest For
Government advisoriesCISA, NCSC, CERT-EU, national CERTsAuthoritative threat alerts, vulnerability warnings
Vendor reportsMandiant, CrowdStrike, Microsoft, Recorded Future, TalosCampaign analysis, actor profiles, malware analysis
Security researchBlogs, conference talks, academic papersNovel techniques, deep dives, vulnerability research
UndergroundForum mirrors, paste sites, Telegram (public), leak aggregatorsThreat actor communications, credential dumps, tools
Code repositoriesGitHub, GitLabMalware source, PoC exploits, attacker tools, leaked configs
Certificate transparencycrt.sh, CensysDomain enumeration, infrastructure mapping
DNS/WHOISSecurityTrails, DomainTools, RiskIQ (via web search)Infrastructure relationships, historical records
News/mediaTech press, infosec news sitesIncident reports, geopolitical context
Social mediaX/Twitter, LinkedIn, Reddit (public)OPSEC failures, community intelligence, actor claims

Search Operator Cheat Sheet

Google Dorks

site:example.com filetype:pdf         # PDFs on a specific site
"threat actor" AND "APT28" -site:reddit.com  # Exclude noise
inurl:"/wp-content/uploads/" filetype:exe    # Exposed uploads
"index of" "/backup"                   # Directory listings
intitle:"login" site:*.example.com     # Login pages on subdomains

GitHub Search

org:target-org password                # Leaked credentials
filename:.env DB_PASSWORD              # Exposed environment files
"api_key" language:python              # Hardcoded API keys
"BEGIN RSA PRIVATE KEY"                # Exposed private keys

Certificate Transparency (crt.sh)

https://crt.sh/?q=%.example.com        # All certs for subdomains
https://crt.sh/?q=example.com&output=json  # JSON output

Shodan Dorks (via web)

ssl.cert.subject.cn:"example.com"      # Certificates for domain
http.title:"Dashboard" org:"Target"    # Web dashboards
product:"Cobalt Strike"                # C2 servers

Collection Process

1. Broad Sweep

Start with broad searches to understand what's available:

  • Search for the topic/actor/indicator across major sources
  • Note the quantity and quality of available information
  • Identify the most promising sources for deeper collection

2. Targeted Collection

Narrow down to specific sources and extract detailed intelligence:

  • Read full reports, not just headlines
  • Extract specific IOCs, TTPs, dates, and attribution claims
  • Note each source's perspective and potential biases

3. Source Assessment

Apply Admiralty Scale to every finding (per source-assessment skill):

  • Rate source reliability (A-F)
  • Rate information credibility (1-6)
  • Note whether information is corroborated across sources

4. Documentation

Record every finding with full attribution:

#### Finding: [Title]
- **Source**: [URL or description]
- **Source rating**: [Admiralty code, e.g., B2]
- **Date collected**: YYYY-MM-DD
- **Date of information**: YYYY-MM-DD (when was this information produced?)
- **Summary**: [Key intelligence extracted]
- **Relevance**: [How this relates to the collection objective]

Ethical and Legal Boundaries

  • Only access publicly available information
  • Do not create fake accounts or impersonate anyone
  • Do not attempt to access restricted systems
  • Respect robots.txt and terms of service
  • Be aware of data protection regulations (GDPR)
  • Do not engage with threat actors or participate in illegal activity

Related skills

When OSINT collection surfaces an indicator, route into the appropriate enrichment / pivot:

  • Bulk IOC enrichment — /ioc-enrichment-workflow chains /lookup-virustotal, /lookup-otx, /lookup-shodan, /lookup-abuseipdb, /lookup-greynoise, /lookup-urlscan, /lookup-censys, and /lookup-misp
  • Single-indicator first hop — /ip-investigation, /domain-investigation, /hash-investigation, /url-investigation
  • Multi-hop graph walk — /indicator-pivoting
  • Internal correlation against your own catalogue — /lookup-misp search-attributes / search-events
  • Ransomware victim-status sweeps on org names — /lookup-ransomwarelive search --q <orgname>
  • Underground forums and Telegram — /darkweb-collection (read its OPSEC primer before any DIY collection)
  • Apply rigor to the finished product — /score-source, /apply-tlp, /confidence-language, /likelihood-language

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.

Keep looking

Skills are one crate of 325,949. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.