transparent-icons

Name: transparent-icons
Author: neuromechanist

37stars7forks

How to install transparent-icons?

neuromechanist/research-skills/transparent-icons

$npx -y skills add neuromechanist/research-skills --skill transparent-icons

Installs into the current project.

Summary

This skill should be used when the user asks to "make an icon", "generate an icon", "create a scientific icon", "make a transparent icon", "make a minimal icon", "icon for a figure", "icon for a paper", "generate an icon set", "batch icons", "icon batch", "batch generate icons",

Files · 1

View on GitHub

SKILL.md

1	# Transparent Icons
2
3	Generate flat scientific icons (Nature/Science journal style) with transparent backgrounds. Used as panel elements by the `[[scientific-figure]]` composer or as standalone graphics for posters, presentations, and grants.
4
5	## Backends
6
7	Two generation backends, auto-selected:
8
9	1. Codex CLI `image_gen` tool (preferred). Uses the local Codex authentication (`~/.codex/auth.json`). No `OPENAI_API_KEY` required. Internally uses gpt-image-2 (April 2026). Default when `codex` is on `$PATH` and authenticated.
10	2. OpenAI Images API (fallback). Calls `client.images.generate(model="gpt-image-2", ...)` directly. Requires `OPENAI_API_KEY` in environment or `.env`.
11
12	Why two backends: many researchers have a ChatGPT subscription and `codex login` but no separate API key. Auto-selection routes through Codex first to keep the no-API-key path working; explicit `--backend codex` or `--backend api` overrides.
13
14	## Transparency
15
16	`gpt-image-2` (April 2026) does not accept `background="transparent"` — both backends always generate opaque PNGs. Transparency is applied as a post-process. Two methods:
17
18	- `threshold` (default) — drop near-white pixels via Pillow. Fast, no extra deps. Works well for flat 2-color icons on a clean white background. Can leave fringes on anti-aliased edges and may erase highlights where the foreground itself is near-white.
19	- `birefnet` (opt-in) — `rembg` + BiRefNet ONNX model with alpha matting. Cleaner edges, handles complex foregrounds. One-time ~400 MB model download on first run; requires extra deps (`rembg`, `onnxruntime`).
20
21	Pick `threshold` for the canonical flat-icon use case. Pick `birefnet` when the icon contains light-colored content (white parts of a brain MRI, light reflections on glass) that the threshold method would erase, or when the icon will be composited at large sizes where fringes are visible.
22
23	## Usage
24
25	Free-form prompt:
26
27	```bash
28	uv run --with openai --with python-dotenv --with pillow python scripts/generate_icon.py \
29	"a human brain with EEG electrodes" -o brain_eeg.png --transparent
30	```
31
32	From the icon bible (curated templates with palettes, descriptions, and prompt hints):
33
34	```bash
35	uv run --with openai --with python-dotenv --with pillow python scripts/generate_icon.py \
36	--template brain-eeg -o brain_eeg.png --transparent
37	```
38
39	With BiRefNet transparency (higher edge quality, extra deps):
40
41	```bash
42	uv run --with openai --with python-dotenv --with pillow --with rembg --with onnxruntime \
43	python scripts/generate_icon.py --template neuron -o neuron.png \
44	--transparent --transparency-method birefnet
45	```
46
47	List templates:
48
49	```bash
50	uv run --with python-dotenv python scripts/generate_icon.py --list-templates
51	```
52
53	Force a specific backend:
54
55	```bash
56	... --backend codex # or --backend api
57	```
58
59	## Theme bible: keep an icon set consistent
60
61	When generating multiple icons that need to look like they belong together (same line weight, same color palette, same perspective), use a `theme.json` file. The schema is in `references/theme.schema.json` and is shared with the `[[ai-full-figure]]` skill. At a minimum:
62
63	```json
64	{
65	"theme_id": "neuro-flat-v1",
66	"palette": {"primary": "#1F3A5F", "accent": "#E07A5F", "neutral": "#F4F1DE", "bg": "transparent"},
67	"stroke": {"weight_px": 4, "linejoin": "round"},
68	"style_tokens": ["flat 2D", "single continuous line", "no shading", "centered"],
69	"negative_tokens": ["text", "labels", "watermark", "gradient", "3D", "shadow"],
70	"composition": {"aspect": "1:1", "padding_pct": 12, "perspective": "orthographic"}
71	}
72	```
73
74	`examples/icon_set.py` demonstrates loading a theme (including stroke weight and line-join hints) and generating an icon set with consistent prompts.
75
76	## When to use this skill vs the icon bible
77
78	- Bible templates (`--template <id>`) are curated for common research subjects (brain, neuron, EEG cap, DNA, microscope, etc.) and produce predictable output. Use them first.
79	- Free-form prompts are right when the subject is not in the bible or needs a specific composition not captured by a template.
80	- `--category` generates every template in a category at once; useful for stocking a project's icon directory.
81
82	See `references/icon-bible.md` for the template schema and `references/prompt-patterns.md` for what works and what

People who liked this also viewed

6 picks

Type

transparent-icons

37stars7forks

View on GitHub

How to install transparent-icons?

neuromechanist/research-skills/transparent-icons

$npx -y skills add neuromechanist/research-skills --skill transparent-icons

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Transparent Icons
2
3	Generate flat scientific icons (Nature/Science journal style) with transparent backgrounds. Used as panel elements by the `[[scientific-figure]]` composer or as standalone graphics for posters, presentations, and grants.
4
5	## Backends
6
7	Two generation backends, auto-selected:
8
9	1. Codex CLI `image_gen` tool (preferred). Uses the local Codex authentication (`~/.codex/auth.json`). No `OPENAI_API_KEY` required. Internally uses gpt-image-2 (April 2026). Default when `codex` is on `$PATH` and authenticated.
10	2. OpenAI Images API (fallback). Calls `client.images.generate(model="gpt-image-2", ...)` directly. Requires `OPENAI_API_KEY` in environment or `.env`.
11
12	Why two backends: many researchers have a ChatGPT subscription and `codex login` but no separate API key. Auto-selection routes through Codex first to keep the no-API-key path working; explicit `--backend codex` or `--backend api` overrides.
13
14	## Transparency
15
16	`gpt-image-2` (April 2026) does not accept `background="transparent"` — both backends always generate opaque PNGs. Transparency is applied as a post-process. Two methods:
17
18	- `threshold` (default) — drop near-white pixels via Pillow. Fast, no extra deps. Works well for flat 2-color icons on a clean white background. Can leave fringes on anti-aliased edges and may erase highlights where the foreground itself is near-white.
19	- `birefnet` (opt-in) — `rembg` + BiRefNet ONNX model with alpha matting. Cleaner edges, handles complex foregrounds. One-time ~400 MB model download on first run; requires extra deps (`rembg`, `onnxruntime`).
20
21	Pick `threshold` for the canonical flat-icon use case. Pick `birefnet` when the icon contains light-colored content (white parts of a brain MRI, light reflections on glass) that the threshold method would erase, or when the icon will be composited at large sizes where fringes are visible.
22
23	## Usage
24
25	Free-form prompt:
26
27	```bash
28	uv run --with openai --with python-dotenv --with pillow python scripts/generate_icon.py \
29	"a human brain with EEG electrodes" -o brain_eeg.png --transparent
30	```
31
32	From the icon bible (curated templates with palettes, descriptions, and prompt hints):
33
34	```bash
35	uv run --with openai --with python-dotenv --with pillow python scripts/generate_icon.py \
36	--template brain-eeg -o brain_eeg.png --transparent
37	```
38
39	With BiRefNet transparency (higher edge quality, extra deps):
40
41	```bash
42	uv run --with openai --with python-dotenv --with pillow --with rembg --with onnxruntime \
43	python scripts/generate_icon.py --template neuron -o neuron.png \
44	--transparent --transparency-method birefnet
45	```
46
47	List templates:
48
49	```bash
50	uv run --with python-dotenv python scripts/generate_icon.py --list-templates
51	```
52
53	Force a specific backend:
54
55	```bash
56	... --backend codex # or --backend api
57	```
58
59	## Theme bible: keep an icon set consistent
60
61	When generating multiple icons that need to look like they belong together (same line weight, same color palette, same perspective), use a `theme.json` file. The schema is in `references/theme.schema.json` and is shared with the `[[ai-full-figure]]` skill. At a minimum:
62
63	```json
64	{
65	"theme_id": "neuro-flat-v1",
66	"palette": {"primary": "#1F3A5F", "accent": "#E07A5F", "neutral": "#F4F1DE", "bg": "transparent"},
67	"stroke": {"weight_px": 4, "linejoin": "round"},
68	"style_tokens": ["flat 2D", "single continuous line", "no shading", "centered"],
69	"negative_tokens": ["text", "labels", "watermark", "gradient", "3D", "shadow"],
70	"composition": {"aspect": "1:1", "padding_pct": 12, "perspective": "orthographic"}
71	}
72	```
73
74	`examples/icon_set.py` demonstrates loading a theme (including stroke weight and line-join hints) and generating an icon set with consistent prompts.
75
76	## When to use this skill vs the icon bible
77
78	- Bible templates (`--template <id>`) are curated for common research subjects (brain, neuron, EEG cap, DNA, microscope, etc.) and produce predictable output. Use them first.
79	- Free-form prompts are right when the subject is not in the bible or needs a specific composition not captured by a template.
80	- `--category` generates every template in a category at once; useful for stocking a project's icon directory.
81
82	See `references/icon-bible.md` for the template schema and `references/prompt-patterns.md` for what works and what

People who liked this also viewed

6 picks

Type