PaperBanana-CN

io.github.mituan-aiPublished by GitHub account mituan-aiv2.0.1Updated Nov 3, 2026

Generate and evaluate academic diagrams and plots with independent model connections.

VerifiedSTDIODesktop onlyData & AnalyticsMedia & Design

Overview

AI-generated overview

Generates academic methodology diagrams and statistical plots from research text or CSV/JSON data, using separately configured VLM and image-generation…

What you can do
  • Create methodology diagrams from method text and a figure caption, with aspect ratio, resolution and output format choices
  • Create statistical plots from CSV or JSON data
  • Evaluate the quality of generated figures and continue a saved run
  • Run full-paper orchestration, batch generation and parameter sweeps
  • Compose multi-panel figures and browse previous runs without any model connection
When to use it
Useful when you are writing a paper or report and want an assistant to draft architecture/methodology figures or data plots, or to batch-generate and evaluate figure variants. It is a desktop-oriented tool, so it suits local research workflows rather than headless server use.
Requirements
Runs locally over stdio via uvx (Python 3.10-3.12 and uv); a desktop browser is needed for the Studio interface. A VLM connection is required for plots and evaluation, and both a VLM and an image-generation connection for methodology diagrams; each connection needs its own protocol, Base URL, API key, model name and timeout. No credentials are declared in the manifest.
Before you install
You must supply API keys for external model services, which are stored in saved connection profiles and used to send your method text, captions and data to those providers. Generation consumes paid provider quota. Editing a saved connection does not activate it, and an empty API-key field keeps the stored key. It is an unofficial community implementation, not affiliated with the upstream project.

Installation

In SourceWeft

  1. Open PaperBanana-CN in the dashboard and add it to a workspace.
  2. Enable the server for the chats that should use its tools.

Desktop only via STDIO. STDIO servers start a local process, so they need the SourceWeft desktop host.

Other MCP clients

Follow the launch instructions in the repository.

README

English · 简体中文

[PaperBanana-CN scientific figure workbench with a generated multimodal fault-diagnosis figure]

PaperBanana-CN generates methodology diagrams from research descriptions and statistical plots from CSV or JSON data. It uses the PaperBanana scientific workflow and adds separate VLM and image-service connections, a Chinese Studio interface, and explicit aspect-ratio and resolution controls.

[Launch Studio] [Install from PyPI]

[CI status] [Package version 2.0.1] [MIT License]

One run from input to figure

[A PaperBanana-CN Studio run progressing from configured inputs to a completed scientific figure]

The recording follows a methodology-diagram run from submitted inputs to the completed result.

Before you start

You need Python 3.10-3.12, uv, and a desktop browser.

TaskRequired model connections
Methodology diagramVLM and image generation
Statistical plotVLM only
Multi-panel composition and run browsingNone

Each connection specifies its own protocol, Base URL, API key, model name, and timeout. The VLM and image roles may use the same service or two different services.

Quick start

1. Launch Studio

bash
uvx paperbanana-cn studio

Open http://127.0.0.1:7860. uvx runs the package in an isolated environment and does not modify Debian or Ubuntu's system Python.

2. Add the model connections

Open Settings → VLM connection, fill in the service fields, and select Save and use. Repeat under Image connection before generating a methodology diagram.

Editing a saved connection does not activate it. An empty API-key field keeps the stored key. Studio does not fill stored keys back into the browser.

Connection manager screenshot and protocol notes

[PaperBanana-CN connection manager with separate fields for a VLM service]

The connection guide lists the supported protocols, credential storage rules, connection tests, and legacy mode.

3. Generate a figure

Open Methodology diagram, provide the method content and figure caption, then choose an aspect ratio, resolution, and output format.

The same task from the CLI:

bash
paperbanana-cn generate \  --input method.txt \  --caption "Overview of the proposed architecture" \  --aspect-ratio 16:9 \  --resolution 2K \  --format png

What PaperBanana-CN adds

Separate VLM and image connections

The two model roles have independent protocol, Base URL, API key, model, and timeout settings. Studio, CLI, and MCP resolve the same saved connections. Saved profiles contain credential references; API keys remain outside the repository and run metadata.

Official APIs, OpenAI-compatible services, and Gemini-compatible services are supported. Provider specifics stay in the adapters rather than the scientific workflow.

Chinese and English Studio

The Studio interface, help text, validation, progress messages, and errors are available in Chinese and English. Changing the interface language does not rewrite prompts, paper text, or labels inside the generated figure.

Aspect ratios and resolution

Supported aspect ratios:

1:1 · 4:3 · 3:2 · 5:4 · 16:9 · 21:9 · 4:5 · 3:4 · 2:3 · 9:16

Resolution tiers:

1K · 2K · 4K

Studio shows the request size or provider-native tier before generation. If an adapter cannot produce the selected combination, validation stops the request and reports the unsupported option.

Studio workflows

AreaWorkflowModel connections
CreateMethodology diagramVLM and image
CreateStatistical plotVLM
ImproveContinue a saved runDepends on the saved run
ImproveQuality evaluationVLM
AutomateFull-paper orchestrationVLM and image
AutomateBatch generationDepends on the task type
AutomateParameter sweepVLM and image
ToolsMulti-panel compositionNone
ToolsRun browserNone

[PaperBanana-CN methodology-diagram workspace with task inputs and a completed result]

Methodology-diagram workspace

[PaperBanana-CN statistical-plot workspace with CSV input and a completed line chart]

Statistical-plot workspace using synthetic demonstration data

CLI, MCP, Docker, and Colab

Entry pointCommand or link
Studiopaperbanana-cn studio
CLIpaperbanana-cn generate --help
MCP serverpaperbanana-cn mcp
GitHub ActionAction reference
Dockerghcr.io/mituan-ai/paperbanana-cn:2.0.1
ColabQuickstart notebook
MCP client configuration
json
{  "mcpServers": {    "paperbanana-cn": {      "command": "uvx",      "args": ["paperbanana-cn", "mcp"]    }  }}

The server provides 11 tools and reads the same active connections as Studio and CLI. See the MCP guide for the tool list and arguments.

Docker
bash
docker run --rm -p 7860:7860 \  -v paperbanana-cn-config:/home/paperbanana/.config/paperbanana-cn \  -v paperbanana-cn-data:/home/paperbanana/.local/share/paperbanana-cn \  -v paperbanana-cn-outputs:/work/outputs \  ghcr.io/mituan-ai/paperbanana-cn:2.0.1 \  studio --host 0.0.0.0
Permanent install, source setup, and optional providers

Install the command in a uv-managed environment:

bash
uv tool install paperbanana-cnpaperbanana-cn studio

Run the current source checkout:

bash
git clone https://github.com/mituan-ai/PaperBanana-CN.gitcd PaperBanana-CNuv syncuv run paperbanana-cn studio

The default package includes Studio, MCP, PDF input, OpenAI-compatible services, and Gemini.

Optional adapterInstall
AWS Bedrockuv tool install "paperbanana-cn[bedrock]"
Anthropicuv tool install "paperbanana-cn[anthropic]"
LiteLLMuv tool install "paperbanana-cn[litellm]"
All optional providersuv tool install "paperbanana-cn[all-providers]"

For CI, use paperbanana-cn connections add --api-key-env ENV_VAR so the key is read from an environment variable instead of a command-line value.

V1 and upstream

V2 is maintained on main as the paperbanana-cn distribution, the paperbanana_cn Python module, and the paperbanana-cn command.

The scientific figure-generation core is based on llmsresearch/paperbanana. PaperBanana-CN is an unofficial community implementation and is not affiliated with or endorsed by the upstream authors.

Community

PaperBanana-CN is maintained by mituan under the MIT License.

Development checks
bash
git clone https://github.com/mituan-ai/PaperBanana-CN.gitcd PaperBanana-CNuv sync --extra devuv run pytest tests/ -quv run ruff check paperbanana_cn/ mcp_server/ tests/ scripts/

Do not upload API keys, private relay URLs, unpublished papers, private datasets, local connection stores, or generated run directories.

Star history

[Star History Chart]

Source: README.md at commit 5f701ec

Tools

0
SourceWeft has not started this package in its sandbox yet, so its tools are not known.

Version history

1
  1. v2.0.1LatestSep 16, 2026

Score

50% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 days25 pts

    Keep the endpoint answering the MCP handshake; SourceWeft probes it every 30 days.

  • Auth declared10 pts

    Declare the credential headers in the registry entry, or keep the endpoint answering the probe so its auth can be classified.

  • Tools listed15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts