hamzaabduljabbar

Construction Drawing Analyzer — Development skill for Claude Code

Development community

Claude skill + Python scripts to analyze construction drawings that automates sheet splitting, polygon extraction, and drawing classification for GCs, estimators, and PMs.

How to install Construction Drawing Analyzer

This entry records only its repository, not the path inside it, so there is no exact command to give. Open hamzaabduljabbar/construction-drawing-analyzer and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Construction Drawing Analyzer does

Claude skill + Python scripts to analyze construction drawings that automates sheet splitting, polygon extraction, and drawing classification for GCs, estimators, and PMs.

Alternatives in Development

Mempalace Mine — Mine projects and conversations into the MemPalace 58.6k ★
Om Tidy — Self-maintenance pass — acts on every hygiene flag: archives completed work, groups loose clusters, splits ove 4.6k ★
Agentsys — This automates everything else · 14 plugins · 43 agents · 30 skills · for Claude Code, OpenCode, Codex 655 ★

README

Drawing Analyzer — Claude AI Skill for Construction Drawings

THIS SKILL AHS BEEN ARCHIVED> KINDLY USE THE OTHER SKILL THA CONVERTS PDF DRAWING IN TO VECTOR DATABASE

Pre-process a set of construction drawing PDFs into structured text indexes so Claude can answer questions **faster, cheaper, and more accurately**.

Built by [Hamza Abdul Jabbar](https://hamzajabbar.online) | AutoConst

The Problem

Raw construction drawing PDFs are terrible AI input.

Claude breaks images into 16×16 pixel tiles — tiny drawing details get lost
A semi-dashed line (cold water) looks identical to a double-dashed line (hot water) at tile resolution
Merged drawing sets regularly fail to upload even under the 30MB limit
Claude Cowork burns 60,000–70,000 tokens per question processing the full PDF set every time
Element counts are approximate — overcounting is common
Cross-references between sheets are not followed

The Solution

Run this skill **once** on a new drawing set. It builds lightweight text indexes. Every future question reads the indexes first, then targets only the specific sheet needed.

	Raw PDF approach	Drawing Analyzer
Tokens per question	60,000–70,000	5,000–8,000
Time per question	5–10 minutes	30–60 seconds
Footing count accuracy	Overcounts	Correct, by type
Area takeoff accuracy	Visual estimate	96% tested
Symbol interpretation	Guesses	Reads actual legend
Cross-references	Not followed	Mapped and followed

Real result

Warehouse slab area on an 11MB drawing set:

Drawing Analyzer: 13,250 m²
Manual Bluebeam measurement: 13,600 m²
Accuracy: 96.7%

What it produces

your-project/
├── indexes/
│   ├── drawings.md              # Master register — loaded on every query (~5KB)
│   ├── sheet-classification.md  # Every sheet classified by type and content
│   ├── cross-reference.md       # Which drawings reference which other drawings
│   └── symbol-librar