doc-extraction

name	doc-extraction
description	Classify and extract structured content from engineering documents using a 3-layer taxonomy: generic content types, engineering patterns, and domain sub-skills. Use when ingesting standards, reports, or technical manuals into structured data for downstream analysis. type: reference
version	1.0.0
updated	"2026-03-12T00:00:00.000Z"
category	engineering
triggers	["document extraction","content classification","extract constants","extract equations","extract tables","parse engineering document","ingest standard","technical manual extraction"]
capabilities	["content-type-classification","engineering-pattern-extraction","standards-reference-parsing","unit-normalization"]
requires	[]
tags	[]
scripts_exempt	true
type	reference

Doc Extraction

When to Use

• Ingesting a new standard or code (DNV-RP, API RP, ISO, ASME)
• Extracting constants, equations, or tables from technical reports
• Building structured datasets from engineering manuals
• Populating knowledge bases from document collections
• Pre-processing documents before analysis workflow

Related Skills

• document-index-pipeline — 7-phase A→G pipeline
• doc-intelligence-promotion — Deep extraction post-processing
• cathodic-protection — CP system design
• viv-analysis — VIV assessment for risers
• fitness-for-service — FFS assessment
• structural-analysis — Structural checks

References

• DNV-RP-B401: Cathodic Protection Design
• DNV-RP-C205: Environmental Conditions and Environmental Loads
• API 579-1/ASME FFS-1: Fitness-for-Service
• API RP 16Q: Design, Selection, Operation, and Maintenance of Marine Drilling Riser Systems

Sub-Skills

• Yield Reality (WRK-1246 Corpus Assessment)
• Architecture
• 1. constants (0% yield — not yet implemented) (+7)
• Unit Detection and Normalization (+4)
• Extraction Workflow
• Output Schema
• Domain Sub-Skills
• Hybrid Classification Strategy (WRK-1188 Learning)