product-spec-pdf-parser

Name: product-spec-pdf-parser
Author: AlpacaLabsLLC

23installs54forks

How to install product-spec-pdf-parser?

alpacalabsllc/skills-for-architects/product-spec-pdf-parser

$npx -y skills add AlpacaLabsLLC/skills-for-architects --skill product-spec-pdf-parser

Installs into the current project.

Summary

Extract structured FF&E product specs from PDF files — price books, fact sheets, and spec sheets — into a standardized schedule. Use when the user shares a product PDF and asks to "parse this price book", extract specs, or import items from a spec sheet.

Files · 1

View on GitHub

SKILL.md

1	# /product-spec-pdf-parser — PDF Product Spec Parser
2
3	Extract structured FF&E data from product PDF files — price books, fact sheets, configurator sheets, and spec sheets. Uses PyMuPDF for text extraction and Claude's reasoning to parse wildly varying PDF layouts into a standardized schedule.
4
5	## Input
6
7	The user provides PDFs in one of these ways:
8
9	1. File paths — one or more PDF file paths
10	2. Folder path — a directory containing PDFs (will process all `.pdf` files)
11	3. Just invoked — ask the user for file paths or a folder
12
13	Also ask (or use defaults):
14
15	- Output destination — Google Sheet, local CSV, or markdown (default: ask)
16	- Variant depth — `expand` (one row per variant/SKU, default) or `summarize` (comma-separated variants in one row)
17
18	## Output Schema
19
20	Products are written to the master Google Sheet — the same 33-column schema used by all product skills, plus PDF-specific extra columns. When writing to CSV, use the same column order.
21
22	Read `../../schema/product-schema.md` (relative to this SKILL.md) for the full column reference, field formats, and category vocabulary. Read `../../schema/sheet-conventions.md` for CRUD patterns with MCP tools.
23
24	Skill-specific column values:
25	- AG (Source): `pdf-parser`
26	- AF (Status): `saved`
27	- J (Link): Blank (no URL for PDFs)
28	- D (Thumbnail): Blank (no image URL typically)
29	- C (Vendor): Blank (source is PDF, not a retailer)
30	- V (Sale Price): Blank (PDFs don't have sale prices)
31	- AC (Image URL): Blank (no image from PDF)
32
33	### PDF-specific data in Notes (col AE)
34
35	PDFs contain fields that don't have dedicated master columns. Append these to Notes using `\|` as delimiter:
36
37	- Variant: `Variant: Diamond, Black`
38	- Price Adder: `Price adder: +$130 (PostureFit SL)`
39	- Country of Origin: `Origin: Sweden`
40	- Source File: `Source: alphabeta-fact-sheet.pdf`
41
42	Example Notes cell: `Variant: Diamond, Black \| Origin: Sweden \| Source: alphabeta-fact-sheet.pdf`
43
44	## Variant Handling
45
46	Different PDF types require different approaches:
47
48	### Fact sheets with SKUs (e.g., Alphabeta lamp)
49	- One row per SKU. Each shade shape × color = one row.
50	- Product Name stays the same across rows. Variant describes the distinguishing attributes.
51	- Example: "Alphabeta Floor Lamp" / Variant: "Diamond, Black" / SKU: "..."
52
53	### Fact sheets with upholstery/finish combos (e.g., Puffy lounge chair)
54	- One row per upholstery option. Frame finish goes in Colors/Finishes.
55	- Distinct products (chair + ottoman) each get their own set of rows.
56	- Example: "Puffy Lounge Chair" / Variant: "Traffic Red" / Colors/Finishes: "Chrome frame"
57
58	### Price books / configurators (e.g., Aeron price book)
59	- One row per distinct product type (e.g., Work Chair, Stool, Side Chair).
60	- Base configuration in main fields. Summarize configuration options — do NOT explode every permutation.
61	- Use Price Adder for incremental costs of add-ons or upgrades.
62	- Example: "Aeron Chair" / Variant: "Size B, Graphite" / List Price: 1395.00 / Price Adder: 130.00 (PostureFit SL)
63
64	### `expand` vs `summarize` mode
65	- expand (default): One row per variant, SKU, or distinct option. Best for procurement and ordering.
66	- summarize: One row per product. Colors/Finishes and Variant are comma-separated lists. Best for quick reference.
67
68	## Workflow
69
70	### Step 1: Get input
71
72	Parse the user's input to identify PDF file(s) and output preferences.
73
74	- If given a folder, list all `.pdf` files and report count
75	- If no PDFs found or path is invalid, ask the user
76	- Confirm variant depth — default to `expand` unless the user says otherwise
77	- Report: "Found N PDF(s) to process."
78
79	### Step 2: Extract text from PDF
80
81	Use PyMuPDF (fitz) to extract text from each PDF. Run this Python script via Bash:
82
83	```python
84	import fitz
85	import sys
86	import json
87
88	pdf_path = sys.argv[1]
89	doc = fitz.open(pdf_path)
90	pages = []
91	for i, page in enumerate(doc):
92	text = page.get_text()
93	pages.append({"page": i + 1, "text": text})
94	doc.close()
95
96	print(json.dumps({"filename": pdf_path.split("/")[-1], "total_pages": len(pages), "pages": pages}))
97	```
98
99	For each PDF, extract all pages and save the JSON output.
100
101	### Step 3: Parse products with Claude
102
103	Read the extracted text and identify all products, variants, and specifications. This is the core intelligence step — Claude reasons over the text to structure it.
104
105	For small PDFs (≤20 pages): Process all pages at once.
106
107	For large PDFs (>20 pages): Process in chunks of 10 pages at a

People who liked this also viewed

6 picks

Type

product-spec-pdf-parser

23installs54forks

View on GitHub

How to install product-spec-pdf-parser?

alpacalabsllc/skills-for-architects/product-spec-pdf-parser

$npx -y skills add AlpacaLabsLLC/skills-for-architects --skill product-spec-pdf-parser

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# /product-spec-pdf-parser — PDF Product Spec Parser
2
3	Extract structured FF&E data from product PDF files — price books, fact sheets, configurator sheets, and spec sheets. Uses PyMuPDF for text extraction and Claude's reasoning to parse wildly varying PDF layouts into a standardized schedule.
4
5	## Input
6
7	The user provides PDFs in one of these ways:
8
9	1. File paths — one or more PDF file paths
10	2. Folder path — a directory containing PDFs (will process all `.pdf` files)
11	3. Just invoked — ask the user for file paths or a folder
12
13	Also ask (or use defaults):
14
15	- Output destination — Google Sheet, local CSV, or markdown (default: ask)
16	- Variant depth — `expand` (one row per variant/SKU, default) or `summarize` (comma-separated variants in one row)
17
18	## Output Schema
19
20	Products are written to the master Google Sheet — the same 33-column schema used by all product skills, plus PDF-specific extra columns. When writing to CSV, use the same column order.
21
22	Read `../../schema/product-schema.md` (relative to this SKILL.md) for the full column reference, field formats, and category vocabulary. Read `../../schema/sheet-conventions.md` for CRUD patterns with MCP tools.
23
24	Skill-specific column values:
25	- AG (Source): `pdf-parser`
26	- AF (Status): `saved`
27	- J (Link): Blank (no URL for PDFs)
28	- D (Thumbnail): Blank (no image URL typically)
29	- C (Vendor): Blank (source is PDF, not a retailer)
30	- V (Sale Price): Blank (PDFs don't have sale prices)
31	- AC (Image URL): Blank (no image from PDF)
32
33	### PDF-specific data in Notes (col AE)
34
35	PDFs contain fields that don't have dedicated master columns. Append these to Notes using `\|` as delimiter:
36
37	- Variant: `Variant: Diamond, Black`
38	- Price Adder: `Price adder: +$130 (PostureFit SL)`
39	- Country of Origin: `Origin: Sweden`
40	- Source File: `Source: alphabeta-fact-sheet.pdf`
41
42	Example Notes cell: `Variant: Diamond, Black \| Origin: Sweden \| Source: alphabeta-fact-sheet.pdf`
43
44	## Variant Handling
45
46	Different PDF types require different approaches:
47
48	### Fact sheets with SKUs (e.g., Alphabeta lamp)
49	- One row per SKU. Each shade shape × color = one row.
50	- Product Name stays the same across rows. Variant describes the distinguishing attributes.
51	- Example: "Alphabeta Floor Lamp" / Variant: "Diamond, Black" / SKU: "..."
52
53	### Fact sheets with upholstery/finish combos (e.g., Puffy lounge chair)
54	- One row per upholstery option. Frame finish goes in Colors/Finishes.
55	- Distinct products (chair + ottoman) each get their own set of rows.
56	- Example: "Puffy Lounge Chair" / Variant: "Traffic Red" / Colors/Finishes: "Chrome frame"
57
58	### Price books / configurators (e.g., Aeron price book)
59	- One row per distinct product type (e.g., Work Chair, Stool, Side Chair).
60	- Base configuration in main fields. Summarize configuration options — do NOT explode every permutation.
61	- Use Price Adder for incremental costs of add-ons or upgrades.
62	- Example: "Aeron Chair" / Variant: "Size B, Graphite" / List Price: 1395.00 / Price Adder: 130.00 (PostureFit SL)
63
64	### `expand` vs `summarize` mode
65	- expand (default): One row per variant, SKU, or distinct option. Best for procurement and ordering.
66	- summarize: One row per product. Colors/Finishes and Variant are comma-separated lists. Best for quick reference.
67
68	## Workflow
69
70	### Step 1: Get input
71
72	Parse the user's input to identify PDF file(s) and output preferences.
73
74	- If given a folder, list all `.pdf` files and report count
75	- If no PDFs found or path is invalid, ask the user
76	- Confirm variant depth — default to `expand` unless the user says otherwise
77	- Report: "Found N PDF(s) to process."
78
79	### Step 2: Extract text from PDF
80
81	Use PyMuPDF (fitz) to extract text from each PDF. Run this Python script via Bash:
82
83	```python
84	import fitz
85	import sys
86	import json
87
88	pdf_path = sys.argv[1]
89	doc = fitz.open(pdf_path)
90	pages = []
91	for i, page in enumerate(doc):
92	text = page.get_text()
93	pages.append({"page": i + 1, "text": text})
94	doc.close()
95
96	print(json.dumps({"filename": pdf_path.split("/")[-1], "total_pages": len(pages), "pages": pages}))
97	```
98
99	For each PDF, extract all pages and save the JSON output.
100
101	### Step 3: Parse products with Claude
102
103	Read the extracted text and identify all products, variants, and specifications. This is the core intelligence step — Claude reasons over the text to structure it.
104
105	For small PDFs (≤20 pages): Process all pages at once.
106
107	For large PDFs (>20 pages): Process in chunks of 10 pages at a

People who liked this also viewed

6 picks

Type