viewport-highlighter

Name: viewport-highlighter
Author: dleerdefi

24stars5forks

How to install viewport-highlighter?

dleerdefi/claude-code-construction/viewport-highlighter

$npx -y skills add dleerdefi/claude-code-construction --skill viewport-highlighter

Installs into the current project.

Summary

Identify and highlight viewports on construction drawing sheets using vision. Detects view boundaries, titles, scales, and view types. Creates viewport overlays via AgentCM API. Requires AgentCM (.construction/ directory). Triggers: 'highlight viewports', 'find views'.

Files · 1

View on GitHub

SKILL.md

1	# Viewport Highlighter
2
3	## Purpose
4
5	Automate viewport segmentation of construction drawing sheets. Each sheet
6	typically contains 1-12+ distinct views (floor plans, sections, elevations,
7	details, schedules). This skill identifies all views, determines their
8	boundaries, extracts metadata (title, detail number, scale, type), and
9	creates viewport highlights — producing the same result as a user manually
10	drawing and labeling each viewport in the UI.
11
12	Does NOT: modify existing viewports, delete viewports, or change element
13	data. Only creates new viewport highlights and populates their metadata.
14
15	---
16
17	## Step 0: Verify AgentCM Mode
18
19	This skill requires AgentCM. It writes viewport overlays through the AgentCM REST API and has no standalone output path.
20
21	Check for `.construction/` directory at the project root.
22
23	If `.construction/` is absent, stop immediately and tell the user:
24	> "viewport-highlighter requires an AgentCM project — `.construction/` directory not found. This skill submits viewport overlays through the AgentCM API and cannot operate without it. Open this project in AgentCM first, then re-run."
25
26	If `.construction/` exists:
27	- Read `.construction/CLAUDE.md` for project context
28	- Read `.construction/database.yaml` for `query_command`, `project_id`, `api_url`
29	- Read `.construction/index/sheet_index.yaml` for sheet inventory
30	- Sheet images at `.construction/rasters/{sheet_number}.png`
31	- OCR data queryable via `extracted_items` table in PostgreSQL
32	- Write viewports via REST API
33
34	## Step 1: User Scopes the Task
35
36	User provides sheet scope:
37	- Specific sheets: "A2.01, A2.02, A5.01"
38	- By discipline: "all architectural sheets"
39	- All sheets: "every sheet in the set"
40
41	Optional: user can specify which view types to look for (e.g., "only floor
42	plans and sections"). Default: identify ALL views on each sheet.
43
44	Before proceeding: Confirm the sheet list and any filters with the user.
45	Show the count of sheets to process.
46
47	## Step 2: Vision — Identify View Boundaries
48
49	For each sheet in scope, rasterize to PNG (if not already available) and
50	examine with vision.
51
52	```bash
53	# Rasterize on demand if the pre-rendered PNG is missing
54	${CLAUDE_SKILL_DIR}/../../bin/construction-python \
55	${CLAUDE_SKILL_DIR}/../../scripts/pdf/rasterize_page.py \
56	"{pdf_path}" {page_index} --dpi 200 --output /tmp/{sheet_number}.png
57	```
58
59	Vision task: Examine the full sheet image. Identify every distinct view
60	(drawing area) on the sheet. For each view, extract:
61
62	\| Field \| What to look for \| Example \|
63	\|-------\|-----------------\|---------\|
64	\| Title \| View title bar text (usually centered below the view) \| "FIRST FLOOR PLAN" \|
65	\| Detail Number \| Number in the detail bubble or title bar \| "1", "A2.01", "3/A5" \|
66	\| Scale \| Scale annotation near the title bar \| `1/8" = 1'-0"`, `1/4" = 1'-0"`, "NTS" \|
67	\| View Type \| Classification of the drawing content \| plan, section, elevation, detail, schedule \|
68	\| Bounding Region \| Approximate rectangle enclosing the entire view (normalized 0-1) \| `{x: 0.02, y: 0.05, w: 0.48, h: 0.65}` \|
69
70	### View boundary detection signals
71
72	Use these visual cues to determine where one view ends and another begins:
73
74	1. Heavy border lines — thick lines separating drawing areas
75	2. Title bars — horizontal bars with view name, scale, detail number
76	3. Detail bubbles — circles or hexagons with detail/sheet reference
77	4. Whitespace gaps — clear separations between drawing content
78	5. Grid systems — column/row grids define the extents of a plan view
79	6. Section cut lines — long dash-dot lines with directional arrows
80	7. Match lines — indicate where a plan continues on another sheet
81
82	### View type classification
83
84	\| View Type \| `extractionScope` value \| Signals \|
85	\|-----------\|------------------------\|---------\|
86	\| Floor plan \| `plan` \| Grid lines, room names/numbers, dimension strings, north arrow \|
87	\| Enlarged plan \| `plan` \| "ENLARGED" in title, larger scale than base plan, room detail \|
88	\| Section \| `section` \| Section cut reference (e.g., "SECTION A-A"), vertical layers, material hatching \|
89	\| Elevation \| `elevation` \| "ELEVATION" in title, facade view, material callouts, floor lines \|
90	\| Detail \| `detail` \| Detail bubble reference, large scale (3"=1'-0"), construction assembly closeup \|
91	\| Schedule \| `schedule` \| Tabular grid, column headers, row data (door schedule, finish schedule) \|
92	\| Diagram \| `other` \| Riser diagrams, single-line diagrams, flow diagrams \|
93
94	### Bounding region estimation from vision
95
96	When estimating the bounding region as normalized 0-1 coordinates:
97
98	- x = left edge of view content / sheet width (0.0 = left edge)
99	- y = top edge of view conte

People who liked this also viewed

6 picks

Type

viewport-highlighter

24stars5forks

View on GitHub

How to install viewport-highlighter?

dleerdefi/claude-code-construction/viewport-highlighter

$npx -y skills add dleerdefi/claude-code-construction --skill viewport-highlighter

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Viewport Highlighter
2
3	## Purpose
4
5	Automate viewport segmentation of construction drawing sheets. Each sheet
6	typically contains 1-12+ distinct views (floor plans, sections, elevations,
7	details, schedules). This skill identifies all views, determines their
8	boundaries, extracts metadata (title, detail number, scale, type), and
9	creates viewport highlights — producing the same result as a user manually
10	drawing and labeling each viewport in the UI.
11
12	Does NOT: modify existing viewports, delete viewports, or change element
13	data. Only creates new viewport highlights and populates their metadata.
14
15	---
16
17	## Step 0: Verify AgentCM Mode
18
19	This skill requires AgentCM. It writes viewport overlays through the AgentCM REST API and has no standalone output path.
20
21	Check for `.construction/` directory at the project root.
22
23	If `.construction/` is absent, stop immediately and tell the user:
24	> "viewport-highlighter requires an AgentCM project — `.construction/` directory not found. This skill submits viewport overlays through the AgentCM API and cannot operate without it. Open this project in AgentCM first, then re-run."
25
26	If `.construction/` exists:
27	- Read `.construction/CLAUDE.md` for project context
28	- Read `.construction/database.yaml` for `query_command`, `project_id`, `api_url`
29	- Read `.construction/index/sheet_index.yaml` for sheet inventory
30	- Sheet images at `.construction/rasters/{sheet_number}.png`
31	- OCR data queryable via `extracted_items` table in PostgreSQL
32	- Write viewports via REST API
33
34	## Step 1: User Scopes the Task
35
36	User provides sheet scope:
37	- Specific sheets: "A2.01, A2.02, A5.01"
38	- By discipline: "all architectural sheets"
39	- All sheets: "every sheet in the set"
40
41	Optional: user can specify which view types to look for (e.g., "only floor
42	plans and sections"). Default: identify ALL views on each sheet.
43
44	Before proceeding: Confirm the sheet list and any filters with the user.
45	Show the count of sheets to process.
46
47	## Step 2: Vision — Identify View Boundaries
48
49	For each sheet in scope, rasterize to PNG (if not already available) and
50	examine with vision.
51
52	```bash
53	# Rasterize on demand if the pre-rendered PNG is missing
54	${CLAUDE_SKILL_DIR}/../../bin/construction-python \
55	${CLAUDE_SKILL_DIR}/../../scripts/pdf/rasterize_page.py \
56	"{pdf_path}" {page_index} --dpi 200 --output /tmp/{sheet_number}.png
57	```
58
59	Vision task: Examine the full sheet image. Identify every distinct view
60	(drawing area) on the sheet. For each view, extract:
61
62	\| Field \| What to look for \| Example \|
63	\|-------\|-----------------\|---------\|
64	\| Title \| View title bar text (usually centered below the view) \| "FIRST FLOOR PLAN" \|
65	\| Detail Number \| Number in the detail bubble or title bar \| "1", "A2.01", "3/A5" \|
66	\| Scale \| Scale annotation near the title bar \| `1/8" = 1'-0"`, `1/4" = 1'-0"`, "NTS" \|
67	\| View Type \| Classification of the drawing content \| plan, section, elevation, detail, schedule \|
68	\| Bounding Region \| Approximate rectangle enclosing the entire view (normalized 0-1) \| `{x: 0.02, y: 0.05, w: 0.48, h: 0.65}` \|
69
70	### View boundary detection signals
71
72	Use these visual cues to determine where one view ends and another begins:
73
74	1. Heavy border lines — thick lines separating drawing areas
75	2. Title bars — horizontal bars with view name, scale, detail number
76	3. Detail bubbles — circles or hexagons with detail/sheet reference
77	4. Whitespace gaps — clear separations between drawing content
78	5. Grid systems — column/row grids define the extents of a plan view
79	6. Section cut lines — long dash-dot lines with directional arrows
80	7. Match lines — indicate where a plan continues on another sheet
81
82	### View type classification
83
84	\| View Type \| `extractionScope` value \| Signals \|
85	\|-----------\|------------------------\|---------\|
86	\| Floor plan \| `plan` \| Grid lines, room names/numbers, dimension strings, north arrow \|
87	\| Enlarged plan \| `plan` \| "ENLARGED" in title, larger scale than base plan, room detail \|
88	\| Section \| `section` \| Section cut reference (e.g., "SECTION A-A"), vertical layers, material hatching \|
89	\| Elevation \| `elevation` \| "ELEVATION" in title, facade view, material callouts, floor lines \|
90	\| Detail \| `detail` \| Detail bubble reference, large scale (3"=1'-0"), construction assembly closeup \|
91	\| Schedule \| `schedule` \| Tabular grid, column headers, row data (door schedule, finish schedule) \|
92	\| Diagram \| `other` \| Riser diagrams, single-line diagrams, flow diagrams \|
93
94	### Bounding region estimation from vision
95
96	When estimating the bounding region as normalized 0-1 coordinates:
97
98	- x = left edge of view content / sheet width (0.0 = left edge)
99	- y = top edge of view conte

People who liked this also viewed

6 picks

Type