DDC Skills for Construction
DDC Skills for AI Agents in Construction
238 skills for construction companies: BIM to Excel export, cost item search, estimates, site reports and document generation
- 375 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Scripts generated from the skills read and write project files: models, estimates, Excel tables and PDFs
- Some skills connect to outside services: Qdrant at the address in `QDRANT_URL`, n8n, the OpenAI API
- The batch conversion skill launches CAD and BIM converter executables
datadrivenconstruction/ddc_skills_for_ai_agents_in_construction
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/ddc-skills-for-ai-agents-in-constructionDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Scripts generated from the skills read and write project files: models, estimates, Excel tables and PDFs
- Some skills connect to outside services: Qdrant at the address in `QDRANT_URL`, n8n, the OpenAI API
- The batch conversion skill launches CAD and BIM converter executables
This is third-party code. Review the repository files before installing.
What it does
DDC Skills is a collection of 238 `SKILL.md` folders for the construction industry. Each skill covers one task, such as moving quantities from an IFC model into Excel, searching a database of cost items, producing a daily site report or comparing budget with actuals, and holds instructions plus a Python code example. You open a skill folder in your assistant, it reads `SKILL.md` and generates code for your data, then you review the result and run it. The skills sit in six folders. `1_DDC_Toolkit` holds 85 ready tools: the CWICR cost database, CAD and BIM converters, analytics. `2_DDC_Book` gathers 67 skills that follow the chapters of the author's book on data in construction. `3_DDC_Insights` has 20 automation scenarios built on n8n, agents and field tools, including a Telegram bot. `4_DDC_Curated` has 20 skills for generating PDF, Excel, DOCX and PPTX documents and for quality checks. `5_DDC_Innovative` has 34 experimental skills: defect detection from photos, digital twins, embodied carbon, EU AI Act requirements. `6_OpenConstructionERP` has 12 skills for working with an open ERP platform for construction. Some skills rely on outside programs and services. For example, `semantic-search-cwicr` searches cost items by meaning in a Qdrant vector store and needs the `QDRANT_URL` variable, while `batch-cad-converter` calls the `RvtExporter`, `DwgExporter` and `DgnExporter` converters and targets Windows.
Who it is for. For estimators, BIM coordinators, project managers and IT staff at construction companies who want an assistant to handle reports, estimates and model processing.
Good fit when
- You need to pull quantities and properties of elements from an IFC or Revit model into an Excel table
- An estimator wants to search cost items by describing the work in their own words, in one of 26 languages
- A project manager wants a daily site report and a regular budget versus actuals comparison
- A company does not know where to start digitalizing and is looking for scattered data sources: the `data-silo-detection` and `data-source-audit` skills cover that
- You need templates for generating PDF, Excel, DOCX and PPTX documents for a project
Not a fit when
- You need a commercial product built on CWICR: the database is released under CC BY-NC 4.0 and the author requires a separate license for commercial use
- You need a ready application with an interface: the skills give instructions and code examples, and you run and refine the result yourself
- You work on macOS or Linux and the skill you need calls the RVT, DWG and DGN converters built for Windows
- The agent should drive an open Revit model directly: the catalog has Revit MCP for that
Example request
Open the ifc-qto-extraction skill and write a script that exports wall and slab quantities from our IFC model to ExcelLimitations
- The skills run nothing themselves: the assistant gets instructions and code examples, so the result has to be checked against your data and refined.
- The MIT license covers the skills' code and instructions. The CWICR cost database lives in a separate repository under CC BY-NC 4.0, and commercial use needs a DDC license. The author's book in the `Books` folder is copyrighted: it is free to download, and redistribution must keep the cover and attribution.
- Most skills need Python 3.9 or newer, and some require Docker for n8n, PostgreSQL or SQLite, or an OpenAI key.
- The author checked the folder structure, the `SKILL.md` format and the data being current. His report does not describe running the skills on specific agents.
- The counts disagree: the repository description says 221 skills, the tables say 238, and the per-folder list adds up to 85, 67, 20, 20, 34 and 12.
How to disable. Delete the cloned `DDC_Skills_for_AI_Agents_in_Construction` folder, or the copied folder of a single skill, from your assistant's skills folder. Files created by the generated code stay wherever you saved them.
FAQ
Which of the six folders should I start with?
The author sorts them by need. If you want a working tool right away, take `1_DDC_Toolkit`. To learn in order, follow the chapters in `2_DDC_Book`. For process automation look at `3_DDC_Insights`, and for documents and reports at `4_DDC_Curated`. `5_DDC_Innovative` is meant for experiments with AI and machine learning, and `6_OpenConstructionERP` for those building on the open ERP platform.
Which Python packages do the skills need?
For the general set the author names `pandas`, `openpyxl`, `ifcopenshell` and `pdfplumber`, installed with `pip install pandas openpyxl ifcopenshell pdfplumber`. The `semantic-search-cwicr` skill also needs `qdrant-client`, `pandas` and `sentence-transformers`.
Can I use the CWICR cost database in a commercial project?
Only under a separate license. The DDC compilation is released under CC BY-NC 4.0, which means non-commercial use. For commercial use the author offers his own license and states in `NOTICE.md` that the database must not be treated as CC BY 4.0. The underlying national data is available from the official issuers on their own terms.
How does cost item search work in semantic-search-cwicr?
The skill encodes a query with the `BAAI/bge-m3` model (1024 dimensions) and looks for the nearest records in Qdrant. Each language has its own collection named like `cwicr_{LANG}_v3`. The data covers 8 national bases with 78,228 items and a global base of 30 markets in 26 languages. On small servers the author uses an ONNX int8 variant of the model of about 700 MB.
What is the claw.json file next to SKILL.md?
It is a manifest for OpenClaw. In the `batch-cad-converter` folder it sets the name, version 2.0.0, MIT license, the `filesystem` permission, the `instructions.md` entry file and minimum OpenClaw version 0.8.0. It is not needed to use a skill through `SKILL.md` in other assistants.
What do the skills in 6_OpenConstructionERP do?
They work with the OpenConstructionERP platform: load cost bases, keep bills of quantities and estimates, take off quantities from BIM, plan in 4D and 5D, use the MCP integration, handle tendering, validation and geo coordination. The folder holds 12 skills.
Discussion
No questions or reviews yet.
Related
Skillsdbt Agent Skills
dbt-agent-skills
Official dbt Labs skills: the agent understands dbt projects, runs workflows and helps with migration
A Google DeepMind skill collection for science tasks: access to biology and chemistry databases, literature search and structural biology
Databricks Field Engineering's agent kit: an installer for Databricks skills, an MCP server and a visual Builder App
A pack of 14 skills for public-data research in Chinese: a lead skill runs the project and 13 specialists cover market, competitors and valuation
High-Stakes Analytics Lab
High-Stakes Analytics & Decision Lab
An evidence-gated analytics skill for high-stakes decisions: profiles data, picks a method the evidence supports, can answer with a data request
Two skills for the CUMCM and MCM/ICM math modeling contests: problem breakdown, model choice, code and the paper
Comparison with similar tools
Entries that share the main search query, and how they differ.
| Entry | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| DDC Skills for Construction (this entry) | no | 375 | ||
| Revit MCP | no | not required | 411 | MCP server for Autodesk Revit: the agent reads the model, creates elements, grids and rooms and calculates material volumes |
| OpenTakeoff | no | not required | 716 | MCP server for construction takeoff |
README in short
The repository description has a table of the six folders with the skill count in each, a list of typical tasks naming the skills (`semantic-search-cwicr`, `estimate-builder`, `n8n-daily-report`), implementation stages from a data audit to machine learning forecasts, and data flow diagrams. After that come the requirements, the folder structure and links to `GETTING_STARTED.md`, `OPTIMIZER_GUIDE.md` and the author's book Data-Driven Construction. Alongside sit `NOTICE.md` with licenses and trademarks, `REVIEW_2026.md` with the results of ten review rounds, and the book itself as a PDF.
README badge
Original SKILL.md
---
name: "semantic-search-cwicr"
description: "Semantic search in the DDC CWICR construction cost database using vector embeddings (BGE-M3, 1024-dim, per-language Qdrant collections). Find similar work items and resources for cost estimation across 8 national bases and 30 markets in 26 languages."
homepage: "https://datadrivenconstruction.io"
metadata: {"openclaw": {"emoji": "🗄️", "os": ["darwin", "linux", "win32"], "homepage": "https://datadrivenconstruction.io", "requires": {"bins": ["python3"], "env": ["QDRANT_URL"]}, "primaryEnv": "QDRANT_URL"}}
---
# Semantic Search in DDC CWICR Database
## Business Case
### Problem Statement
Construction cost estimation requires finding relevant work items from large databases. Traditional keyword search fails when:
- Users describe work in natural language
- Terminology varies across regions and languages
- Similar work items have different naming conventions
### Solution
DDC CWICR provides pre-computed embeddings (BAAI/bge-m3, 1024 dimensions) enabling multilingual semantic search across **8 national bases (78,228 positions)** plus the **30-market global base** in **26 languages**, with 48 PPP-repriced market catalogs per national base.
### Business Value
- **90% faster** work item lookup compared to manual search
- **Multi-language**: Arabic, Bulgarian, Chinese, Croatian, Czech, Danish, Dutch, English, Finnish, French, German, Hindi, Indonesian, Italian, Japanese, Korean, Mongolian, Norwegian, Polish, Portuguese, Romanian, Russian, Spanish, Swedish, Thai, Turkish, Vietnamese
- **Higher accuracy** by finding semantically similar items, not just keyword matches
## Data landscape (2026)
| National base | Region id | Positions |
|---|---|---|
| Turkey (Birim Fiyat) | `TR_NATIONAL` | 22,704 |
| China (Beijing Dinge + Bole) | `ZH_CHINA` | 11,312 |
| Brazil (SINAPI) | `BR_NATIONAL` | 9,723 |
| Spain (BCCA Andalucía) | `ES_ANDALUCIA` | 6,453 |
| Italy (Prezzario Toscana) | `IT_TOSCANA` | 5,836 |
| Vietnam (Dinh Muc) | `VN_NATIONAL` | 4,299 |
| Indonesia (AHSP) | `ID_NATIONAL` | 2,784 |
| Greece (GGDE) | `GR_NATIONAL` | 2,647 |
Each base ships the **95-column CWICR master schema** (`rate_code`, `rate_original_name`, `rate_final_name`, `rate_unit`, `total_cost_per_position`, classification hierarchy `collection/department/section/subsection/category`, `resource_*` component lines with `is_material/is_machine/is_labor` flags) plus 26 language editions and 48 `markets/*.csv` catalogs.
Latest data release: **v0.4.0** (see [releases](https://github.com/datadrivenconstruction/OpenConstructionEstimate-DDC-CWICR/releases)).
## Technical Implementation
### Prerequisites