DDC_Skills_for_AI_Agents_in_Construction MCP — 사용 사례, 설치 & 라이브 데모

왜 쓰나요

핵심 기능

Breadth: 221 sub-skills across the construction lifecycle
BIM data handling (IFC parsing, Revit exports, model audits)
Cost estimation templates and WBS conventions
Scheduling helpers (P6/MS Project-adjacent)
Document control: submittals, RFIs, change orders

라이브 데모

실제 사용 모습

ddc-skills-for-ai-agents-in-construction-skill.replay ▶ 준비됨

0/0

설치

클라이언트 선택

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

{
  "mcpServers": {
    "ddc-skills-for-ai-agents-in-construction-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/datadrivenconstruction/DDC_Skills_for_AI_Agents_in_Construction",
        "~/.claude/skills/DDC_Skills_for_AI_Agents_in_Construction"
      ],
      "_inferred": true
    }
  }
}

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

~/.cursor/mcp.json · .cursor/mcp.json

{
  "mcpServers": {
    "ddc-skills-for-ai-agents-in-construction-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/datadrivenconstruction/DDC_Skills_for_AI_Agents_in_Construction",
        "~/.claude/skills/DDC_Skills_for_AI_Agents_in_Construction"
      ],
      "_inferred": true
    }
  }
}

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

VS Code → Cline → MCP Servers → Edit

{
  "mcpServers": {
    "ddc-skills-for-ai-agents-in-construction-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/datadrivenconstruction/DDC_Skills_for_AI_Agents_in_Construction",
        "~/.claude/skills/DDC_Skills_for_AI_Agents_in_Construction"
      ],
      "_inferred": true
    }
  }
}

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

~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "ddc-skills-for-ai-agents-in-construction-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/datadrivenconstruction/DDC_Skills_for_AI_Agents_in_Construction",
        "~/.claude/skills/DDC_Skills_for_AI_Agents_in_Construction"
      ],
      "_inferred": true
    }
  }
}

Claude Desktop과 같은 형식. Windsurf 재시작 후 적용.

~/.continue/config.json

{
  "mcpServers": [
    {
      "name": "ddc-skills-for-ai-agents-in-construction-skill",
      "command": "git",
      "args": [
        "clone",
        "https://github.com/datadrivenconstruction/DDC_Skills_for_AI_Agents_in_Construction",
        "~/.claude/skills/DDC_Skills_for_AI_Agents_in_Construction"
      ]
    }
  ]
}

Continue는 맵이 아닌 서버 오브젝트 배열 사용.

~/.config/zed/settings.json

{
  "context_servers": {
    "ddc-skills-for-ai-agents-in-construction-skill": {
      "command": {
        "path": "git",
        "args": [
          "clone",
          "https://github.com/datadrivenconstruction/DDC_Skills_for_AI_Agents_in_Construction",
          "~/.claude/skills/DDC_Skills_for_AI_Agents_in_Construction"
        ]
      }
    }
  }
}

context_servers에 추가. 저장 시 Zed가 핫 리로드.

claude mcp add ddc-skills-for-ai-agents-in-construction-skill -- git clone https://github.com/datadrivenconstruction/DDC_Skills_for_AI_Agents_in_Construction ~/.claude/skills/DDC_Skills_for_AI_Agents_in_Construction

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

사용 사례

실전 활용법: DDC_Skills_for_AI_Agents_in_Construction

How to audit a BIM IFC model for basic data quality

👤 BIM coordinators and VDC teams ⏱ ~60 min advanced

언제 쓸까: You receive an IFC file from a consultant and need a quality check.

사전 조건

Python with ifcopenshell — pip install ifcopenshell
Skill cloned — git clone https://github.com/datadrivenconstruction/DDC_Skills_for_AI_Agents_in_Construction ~/.claude/skills/DDC_Skills_for_AI_Agents_in_Construction

흐름

Parse the IFC

Audit model.ifc — count entities by class, find missing Psets, flag orphan geometry.✓ 복사됨

→ Entity counts + quality flags
Check naming and classification

Verify wall and room naming against our company standard.✓ 복사됨

→ Per-element conformance report
Produce a coordinator-ready report

Generate an issue list the consultant can act on.✓ 복사됨

→ Ordered issue list

결과: A model audit report you can send back in under an hour.

함정

Huge IFC files exceed memory — Stream-parse rather than load full tree; or pre-filter by spatial structure

함께 쓰기: filesystem

Produce a quick order-of-magnitude cost estimate

👤 Estimators, PMs, bid teams ⏱ ~30 min intermediate

언제 쓸까: Early-phase numbers for a concept before you commit to a detailed estimate.

흐름

Describe the project

Concrete frame office, ~12,000 sqm, Tier-1 city, shell-only. Give me a per-sqm cost range.✓ 복사됨

→ Range with the assumptions stated
Break into WBS

Break the estimate into WBS sections with % allocations.✓ 복사됨

→ Top-level WBS with line items
Flag risks

List the top 5 cost risks for this concept.✓ 복사됨

→ Risks with impact

결과: A defensible concept estimate ready for internal review.

함정

Using global averages without local adjustment — Always override with local benchmarks when you have them

Review a construction schedule for common issues

👤 Schedulers and PMs ⏱ ~30 min intermediate

언제 쓸까: You got a schedule from a subcontractor and need to check for the obvious red flags.

흐름

Feed the schedule

Review schedule.xlsx — flag missing logic, negative float, open ends, and unrealistic durations.✓ 복사됨

→ Categorized findings
Prioritize

Rank by impact on completion date.✓ 복사됨

→ Sorted list

결과: Quick surface-level schedule review that catches 80% of common problems.

함정

Missing CPM-level analysis — For real float analysis, export from P6 rather than relying on Excel

조합

다른 MCP와 조합해 10배 효율

ddc-skills-for-ai-agents-in-construction-skill + filesystem

Process a whole project folder of IFCs, schedules, and docs

Walk project/ and produce a weekly coordination packet.✓ 복사됨

도구

이 MCP가 노출하는 것

도구	입력	언제 호출	비용
bim-audit	IFC path	On any incoming IFC	ifcopenshell compute
cost-estimation	project concept	Early phase	0
schedule-review	schedule export	Sub-submittal review	0
document-control	RFIs/submittals/COs	Project admin	0

비용 및 제한

운영 비용

API 쿼터: none
호출당 토큰: 5–30k per task
금액: free at skill level
팁: Use only the sub-skills relevant to your role — this bundle is wide

보안

권한, 시크릿, 파급범위

자격 증명 저장: none

데이터 외부 송신: none

Construction estimates and schedules are often confidential — handle like other sensitive project data

문제 해결

자주 발생하는 오류와 해결

ifcopenshell install fails

Use Python 3.10 or 3.11 and pip install ifcopenshell from a wheel matching your platform.

확인: python -c 'import ifcopenshell; print(ifcopenshell.version)'

Estimates are wildly off

Override unit costs with your regional database; the skill uses generic benchmarks only

대안

DDC_Skills_for_AI_Agents_in_Construction 다른 것과 비교

대안	언제 쓰나	단점/장점
generic csv-data-summarizer-claude-skill	Your construction data is CSV and you don't need BIM semantics	No domain knowledge

DDC_Skills_for_AI_Agents_in_Construction