cti-expert MCP — 사용 사례, 설치 & 라이브 데모

왜 쓰나요

핵심 기능

67+ command shortcuts for common CTI tasks
35 analytic techniques (ACH, diamond model, SATs)
MITRE ATT&CK mapping helpers
Public-OSINT only — no API keys, no paid services required
Structured reporting templates (executive, tactical, technical)

라이브 데모

실제 사용 모습

cti-expert-skill.replay ▶ 준비됨

0/0

설치

클라이언트 선택

~/Library/Application Support/Claude/claude_desktop_config.json · Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "cti-expert-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/7onez/cti-expert",
        "~/.claude/skills/cti-expert"
      ],
      "_inferred": true
    }
  }
}

Claude Desktop → Settings → Developer → Edit Config 열기. 저장 후 앱 재시작.

~/.cursor/mcp.json · .cursor/mcp.json

{
  "mcpServers": {
    "cti-expert-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/7onez/cti-expert",
        "~/.claude/skills/cti-expert"
      ],
      "_inferred": true
    }
  }
}

Cursor는 Claude Desktop과 동일한 mcpServers 스키마 사용. 프로젝트 설정이 전역보다 우선.

VS Code → Cline → MCP Servers → Edit

{
  "mcpServers": {
    "cti-expert-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/7onez/cti-expert",
        "~/.claude/skills/cti-expert"
      ],
      "_inferred": true
    }
  }
}

Cline 사이드바의 MCP Servers 아이콘 클릭 후 "Edit Configuration" 선택.

~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "cti-expert-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/7onez/cti-expert",
        "~/.claude/skills/cti-expert"
      ],
      "_inferred": true
    }
  }
}

Claude Desktop과 같은 형식. Windsurf 재시작 후 적용.

~/.continue/config.json

{
  "mcpServers": [
    {
      "name": "cti-expert-skill",
      "command": "git",
      "args": [
        "clone",
        "https://github.com/7onez/cti-expert",
        "~/.claude/skills/cti-expert"
      ]
    }
  ]
}

Continue는 맵이 아닌 서버 오브젝트 배열 사용.

~/.config/zed/settings.json

{
  "context_servers": {
    "cti-expert-skill": {
      "command": {
        "path": "git",
        "args": [
          "clone",
          "https://github.com/7onez/cti-expert",
          "~/.claude/skills/cti-expert"
        ]
      }
    }
  }
}

context_servers에 추가. 저장 시 Zed가 핫 리로드.

claude mcp add cti-expert-skill -- git clone https://github.com/7onez/cti-expert ~/.claude/skills/cti-expert

한 줄 명령. claude mcp list로 확인, claude mcp remove로 제거.

사용 사례

실전 활용법: cti-expert

How to enrich a suspicious indicator (IP, domain, hash) with public OSINT

👤 SOC analysts, incident responders, blue-team engineers ⏱ ~15 min intermediate

언제 쓸까: You have an IOC from an alert and need context before escalating.

사전 조건

Skill cloned — git clone https://github.com/7onez/cti-expert ~/.claude/skills/cti-expert

흐름

Submit the indicator

Enrich 185.234.218.95 using public OSINT — WHOIS, passive DNS (free sources), reputation feeds.✓ 복사됨

→ Structured context: ASN, geo, historical resolutions, known bad
Map to MITRE ATT&CK if applicable

If this matches a known campaign, map to ATT&CK tactics/techniques.✓ 복사됨

→ TTP list with ATT&CK IDs
Write up

Produce a tactical report: what it is, confidence level, recommended actions.✓ 복사됨

→ Short, decision-oriented write-up

결과: A defensible enrichment you can attach to a ticket in minutes.

함정

Treating reputation scores as ground truth — Record the source + date; note confidence explicitly
OSINT queries accidentally tip off the target — Only use passive sources; no active scanning

함께 쓰기: filesystem

Draft a campaign write-up from a handful of artifacts

👤 CTI analysts producing internal reports ⏱ ~45 min advanced

언제 쓸까: You've been asked to turn a pile of IOCs, TTPs, and notes into a readable brief.

흐름

Collate the artifacts

Here are the IOCs and notes. Cluster them by diamond-model facet (adversary, capability, infra, victim).✓ 복사됨

→ Diamond-model-shaped clustering
Map to ATT&CK

Map observed behaviors to ATT&CK; note coverage gaps.✓ 복사됨

→ Coverage matrix
Write the brief

Produce an executive section, a tactical section, and an IOC appendix.✓ 복사됨

→ 3-part report

결과: A consistent, audience-tiered report ready for review.

함정

Inflating confidence language ('almost certainly') — Force IC-style qualifiers (high/moderate/low) with reasoning

Run an ACH (Analysis of Competing Hypotheses) walk-through

👤 Analysts wrestling with ambiguous attribution ⏱ ~30 min advanced

언제 쓸까: Multiple actors could explain the evidence; you want structure before picking one.

흐름

List hypotheses + evidence

Hypotheses: H1, H2, H3. Evidence: E1–E10. Set up the ACH matrix.✓ 복사됨

→ Matrix with consistency marks per cell
Identify diagnostic evidence

Which evidence items have the highest diagnostic value?✓ 복사됨

→ Evidence ranked by ability to discriminate
Report the least-inconsistent hypothesis

Pick the least-inconsistent hypothesis and list what would flip it.✓ 복사됨

→ Selection + counter-evidence triggers

결과: A transparent attribution decision with documented reasoning.

함정

Anchoring on the first hypothesis — Force the matrix first, conclusion last

조합

다른 MCP와 조합해 10배 효율

cti-expert-skill + filesystem

Process a directory of raw alerts into structured findings

Enrich every IOC under alerts/2025-04-01/ and produce a consolidated daily tactical report.✓ 복사됨

도구

이 MCP가 노출하는 것

도구	입력	언제 호출	비용
indicator-enrichment	IOC (IP/domain/hash/URL)	Triage of any incoming indicator	free OSINT queries
attck-mapping	observed behaviors	When writing up campaigns or detections	0
structured-analytic-techniques	hypotheses + evidence	Ambiguous analysis problems	0
report-templates	findings	Final reporting step	0

비용 및 제한

운영 비용

API 쿼터: depends on the OSINT sources used; all default sources are free-tier
호출당 토큰: 2–10k tokens per indicator workflow
금액: free
팁: Start with passive sources only; add paid feeds only if coverage gaps justify it

보안

권한, 시크릿, 파급범위

자격 증명 저장: none by default; any paid feed API key lives in env vars

데이터 외부 송신: only the OSINT endpoints you allow-list

Strictly defensive use — do not use enrichment output to target individuals or organizations
OSINT queries may be logged by target infrastructure; prefer passive sources
Respect legal frameworks (GDPR, data protection) when handling IOCs tied to persons

문제 해결

자주 발생하는 오류와 해결

Rate limited by a public source

Throttle or switch to an alternate source — many OSINT endpoints are sensitive

Claude hallucinates an enrichment result

Require citations for every claim; reject unsourced enrichment

ATT&CK mapping is too generic

Provide concrete behaviors (processes, commands, network patterns), not high-level summaries

대안

cti-expert 다른 것과 비교

대안	언제 쓰나	단점/장점
claude-cybersecurity-skill	You want broader security skills, not just CTI	Wider but shallower in CTI specifics
claude-security-research-skill	You're researching vulnerabilities rather than triaging indicators	Different job entirely

cti-expert