paperbanana

Name: paperbanana
Author: PlutoLei

37stars3forks

How to install paperbanana?

plutolei/paperbanana-skill/paperbanana

$npx -y skills add PlutoLei/paperbanana-skill --skill paperbanana

Installs into the current project.

Summary

Use when user needs academic diagrams, methodology figures, statistical plots, or presentation slides from text descriptions or data files. Also use for evaluating generated figures against references.

Files · 1

View on GitHub

SKILL.md

1	# PaperBanana - Academic Illustration Generator
2
3	Multi-agent pipeline (Retriever → Planner → Stylist → Visualizer → Critic) for publication-quality academic diagrams, statistical plots, and presentation slides.
4
5	API key: Set provider keys in PaperBanana project's `.env` file.
6	Timeout: 300000 (5 min) for all generation commands.
7
8	---
9
10	## Commands
11
12	All commands run from project root: `cd <paperbanana_dir> && python -m paperbanana.cli <cmd>`
13
14	### Command Selection Decision Tree
15
16	Route user requests to the right subcommand before looking up parameters:
17
18	\| User intent \| Signal words \| Subcommand \|
19	\|-------------\|--------------\|------------\|
20	\| 方法论/架构/流程图 from text or PDF \| "method figure", "架构图", "流程图", "methodology", "pipeline diagram", "论文配图" \| `generate` \|
21	\| Statistical plot from data file \| "plot", "curve", "bar chart", "scatter", "heatmap", has CSV/JSON \| `plot` \|
22	\| Single presentation slide \| "slide", "一张幻灯片", "封面图", single prompt file \| `slide` \|
23	\| Batch slide generation \| "all slides", "批量生成", "N 张幻灯片", `prompts/` directory \| `slide-batch` \|
24	\| Compare generated vs human reference \| "evaluate", "对比", "与参考图对比" \| `evaluate` \|
25	\| Manage reference dataset \| "download dataset", "清缓存" \| `data` \|
26	\| First-time provider config \| "setup", "配置 API key" \| `setup` \|
27
28	Ambiguous input: If user provides just a description with no subcommand signal, default to `generate` (see Argument Parsing table for details).
29
30	Out-of-scope: Pure code generation (matplotlib/seaborn script) is NOT paperbanana's job — those go to `matplotlib` / `scientific-visualization` skills. Paperbanana is for AI-driven image generation + critique loops.
31
32	> Note (upstream sync pending): Upstream `paperbanana` CLI also adds subcommands (`plot-batch` #123, `sweep` #118) not yet reflected in this table. See the [llmsresearch/paperbanana CHANGELOG](https://github.com/llmsresearch/paperbanana) for the authoritative CLI surface.
33
34	### `generate` — Methodology Diagrams
35
36	```bash
37	python -m paperbanana.cli generate --input '<file>' --caption '<caption>' --optimize --verbose
38	```
39
40	When user provides inline text (no file): write to temp file, use as `--input`.
41
42	\| Parameter \| Default \| Description \|
43	\|-----------\|---------\|-------------\|
44	\| `--input` / `-i` \| — \| Path to methodology text file or PDF (`.pdf` requires `pip install 'paperbanana'`) \|
45	\| `--caption` / `-c` \| — \| Figure caption / communicative intent \|
46	\| `--output` / `-o` \| auto \| Output image path \|
47	\| `--vlm-provider` \| `gemini` \| VLM provider: `gemini`, `anthropic`, `openai`, `bedrock`, `openrouter`, `ollama`, `claude_code`, `litellm` \|
48	\| `--vlm-model` \| auto \| VLM model name \|
49	\| `--image-provider` \| auto \| Image gen provider: `google_imagen`, `openai`, `bedrock`, `openrouter` \|
50	\| `--image-model` \| auto \| Image gen model name \|
51	\| `--iterations` / `-n` \| `3` \| Max critic rounds \|
52	\| `--auto` \| off \| Loop until critic is satisfied (safety cap via `--max-iterations`) \|
53	\| `--max-iterations` \| `30` \| Safety cap for `--auto` mode \|
54	\| `--optimize` \| off \| Preprocess inputs (parallel enrichment + caption sharpening) \|
55	\| `--continue` \| off \| Continue from the latest run \|
56	\| `--continue-run` \| — \| Continue from a specific run ID \|
57	\| `--feedback` \| — \| User feedback for the critic when continuing a run \|
58	\| `--aspect-ratio` / `-ar` \| auto \| Target aspect ratio: `1:1`, `2:3`, `3:2`, `3:4`, `4:3`, `9:16`, `16:9`, `21:9` \|
59	\| `--format` / `-f` \| `png` \| Output format: `png`, `jpeg`, `webp` \|
60	\| `--dry-run` \| off \| Validate inputs without making API calls \|
61	\| `--exemplar-retrieval` \| off \| Enable external exemplar retrieval before planning \|
62	\| `--seed` \| — \| Random seed for reproducible generation \|
63	\| `--verbose` / `-v` \| off \| Show detailed agent progress and timing \|
64	\| `--auto-download-data` \| off \| Auto-download expanded reference set (~257MB) on first run \|
65	\| `--venue` \| — \| Academic venue style: `neurips`, `icml`, `acl`, `ieee`, `custom` \|
66	\| `--pages` \| — \| Page range for PDF input (e.g., `3-5`) \|
67	\| `--config` \| — \| Path to config YAML file \|
68
69	> Venue styles: `--venue neurips` applies NeurIPS-specific methodology and plot style guides from `data/guidelines/`. Each venue has distinct color palettes, layout conventions, and typography expectations.
70
71	> PDF input: `--input paper.pdf --pages 3-5` extracts text from the specified pages as source context.
72
73	> Exemplar advanced flags: `--exemplar-retrieval` enables retrieval; see `generate --help` for additional config flags (`--exemplar-endpoint`, `--exemplar-mode`, `--exemplar-top-k`, `--exemplar-timeout`, `--exemplar-retries`).
74
75	### `plot` — Statisti

People who liked this also viewed

6 picks

Type

paperbanana

37stars3forks

View on GitHub

How to install paperbanana?

plutolei/paperbanana-skill/paperbanana

$npx -y skills add PlutoLei/paperbanana-skill --skill paperbanana

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# PaperBanana - Academic Illustration Generator
2
3	Multi-agent pipeline (Retriever → Planner → Stylist → Visualizer → Critic) for publication-quality academic diagrams, statistical plots, and presentation slides.
4
5	API key: Set provider keys in PaperBanana project's `.env` file.
6	Timeout: 300000 (5 min) for all generation commands.
7
8	---
9
10	## Commands
11
12	All commands run from project root: `cd <paperbanana_dir> && python -m paperbanana.cli <cmd>`
13
14	### Command Selection Decision Tree
15
16	Route user requests to the right subcommand before looking up parameters:
17
18	\| User intent \| Signal words \| Subcommand \|
19	\|-------------\|--------------\|------------\|
20	\| 方法论/架构/流程图 from text or PDF \| "method figure", "架构图", "流程图", "methodology", "pipeline diagram", "论文配图" \| `generate` \|
21	\| Statistical plot from data file \| "plot", "curve", "bar chart", "scatter", "heatmap", has CSV/JSON \| `plot` \|
22	\| Single presentation slide \| "slide", "一张幻灯片", "封面图", single prompt file \| `slide` \|
23	\| Batch slide generation \| "all slides", "批量生成", "N 张幻灯片", `prompts/` directory \| `slide-batch` \|
24	\| Compare generated vs human reference \| "evaluate", "对比", "与参考图对比" \| `evaluate` \|
25	\| Manage reference dataset \| "download dataset", "清缓存" \| `data` \|
26	\| First-time provider config \| "setup", "配置 API key" \| `setup` \|
27
28	Ambiguous input: If user provides just a description with no subcommand signal, default to `generate` (see Argument Parsing table for details).
29
30	Out-of-scope: Pure code generation (matplotlib/seaborn script) is NOT paperbanana's job — those go to `matplotlib` / `scientific-visualization` skills. Paperbanana is for AI-driven image generation + critique loops.
31
32	> Note (upstream sync pending): Upstream `paperbanana` CLI also adds subcommands (`plot-batch` #123, `sweep` #118) not yet reflected in this table. See the [llmsresearch/paperbanana CHANGELOG](https://github.com/llmsresearch/paperbanana) for the authoritative CLI surface.
33
34	### `generate` — Methodology Diagrams
35
36	```bash
37	python -m paperbanana.cli generate --input '<file>' --caption '<caption>' --optimize --verbose
38	```
39
40	When user provides inline text (no file): write to temp file, use as `--input`.
41
42	\| Parameter \| Default \| Description \|
43	\|-----------\|---------\|-------------\|
44	\| `--input` / `-i` \| — \| Path to methodology text file or PDF (`.pdf` requires `pip install 'paperbanana'`) \|
45	\| `--caption` / `-c` \| — \| Figure caption / communicative intent \|
46	\| `--output` / `-o` \| auto \| Output image path \|
47	\| `--vlm-provider` \| `gemini` \| VLM provider: `gemini`, `anthropic`, `openai`, `bedrock`, `openrouter`, `ollama`, `claude_code`, `litellm` \|
48	\| `--vlm-model` \| auto \| VLM model name \|
49	\| `--image-provider` \| auto \| Image gen provider: `google_imagen`, `openai`, `bedrock`, `openrouter` \|
50	\| `--image-model` \| auto \| Image gen model name \|
51	\| `--iterations` / `-n` \| `3` \| Max critic rounds \|
52	\| `--auto` \| off \| Loop until critic is satisfied (safety cap via `--max-iterations`) \|
53	\| `--max-iterations` \| `30` \| Safety cap for `--auto` mode \|
54	\| `--optimize` \| off \| Preprocess inputs (parallel enrichment + caption sharpening) \|
55	\| `--continue` \| off \| Continue from the latest run \|
56	\| `--continue-run` \| — \| Continue from a specific run ID \|
57	\| `--feedback` \| — \| User feedback for the critic when continuing a run \|
58	\| `--aspect-ratio` / `-ar` \| auto \| Target aspect ratio: `1:1`, `2:3`, `3:2`, `3:4`, `4:3`, `9:16`, `16:9`, `21:9` \|
59	\| `--format` / `-f` \| `png` \| Output format: `png`, `jpeg`, `webp` \|
60	\| `--dry-run` \| off \| Validate inputs without making API calls \|
61	\| `--exemplar-retrieval` \| off \| Enable external exemplar retrieval before planning \|
62	\| `--seed` \| — \| Random seed for reproducible generation \|
63	\| `--verbose` / `-v` \| off \| Show detailed agent progress and timing \|
64	\| `--auto-download-data` \| off \| Auto-download expanded reference set (~257MB) on first run \|
65	\| `--venue` \| — \| Academic venue style: `neurips`, `icml`, `acl`, `ieee`, `custom` \|
66	\| `--pages` \| — \| Page range for PDF input (e.g., `3-5`) \|
67	\| `--config` \| — \| Path to config YAML file \|
68
69	> Venue styles: `--venue neurips` applies NeurIPS-specific methodology and plot style guides from `data/guidelines/`. Each venue has distinct color palettes, layout conventions, and typography expectations.
70
71	> PDF input: `--input paper.pdf --pages 3-5` extracts text from the specified pages as source context.
72
73	> Exemplar advanced flags: `--exemplar-retrieval` enables retrieval; see `generate --help` for additional config flags (`--exemplar-endpoint`, `--exemplar-mode`, `--exemplar-top-k`, `--exemplar-timeout`, `--exemplar-retries`).
74
75	### `plot` — Statisti

People who liked this also viewed

6 picks

Type