Log inSubmit server
PaperBanana logo

PaperBanana

Generate academic diagrams and statistical plots from text through multi-agent AI workflows and supported model providers.

InstallRepositoryWebsite
Full README Compare alternatives Embed widget Claim this listing

Install PaperBanana

Pick your client, then add it in one click or paste the config.

STDIOPython
~/Library/Application Support/Claude/claude_desktop_config.json
claude_desktop_config.json
{
  "mcpServers": {
    "paperbanana": {
      "command": "uvx",
      "args": [
        "paperbanana"
      ],
      "env": {
        "OPENAI_API_KEY": "YOUR_VALUE_HERE",
        "GOOGLE_API_KEY": "YOUR_VALUE_HERE",
        "OPENAI_BASE_URL": "YOUR_VALUE_HERE",
        "GOOGLE_BASE_URL": "YOUR_VALUE_HERE",
        "GOOGLE_VLM_MODEL": "YOUR_VALUE_HERE",
        "GOOGLE_IMAGE_MODEL": "YOUR_VALUE_HERE"
      }
    }
  }
}

Settings → Developer → Edit Config, merge this into mcpServers, then fully quit and reopen Claude Desktop.

Values are written into the snippet above so you can paste it as is. They never leave your browser.

Install not yet sandbox-tested— we haven't run this command through our automated sandbox yet. Not a red flag; we're working through the catalog.

Overview

PaperBanana MCP server generates academic diagrams and statistical plots from text descriptions through a multi-agent pipeline. It can use OpenAI, Azure OpenAI or Foundry, Google Gemini, and Atlas Cloud providers, with optional input optimization and iterative refinement. Reach for it when an AI agent needs to turn methodology text into publication-oriented visual outputs or create plots from structured data. The project also provides CLI, Python, local Studio, batch, and PDF-context workflows.

Use cases

  • Generate a methodology diagram from a research description
  • Create statistical plots from CSV or JSON batch inputs
  • Refine an academic illustration with iterative feedback
  • Add selected PDF pages as context for diagram generation

Key features

  • Multi-agent diagram and plot generation
  • OpenAI, Azure, Gemini, and Atlas Cloud provider support
  • Input optimization and automatic refinement
  • Batch diagram and plot generation
  • Optional PDF context
  • CLI, Python API, Studio, and MCP access

Security & health

What it can reach and how reliably it runs — from our own checks, not the maintainer’s claims.

ActiveLast checked 2d ago

Recent health check succeeded.

12/12 checks healthy over the last 45d

Supply-chain signal

No high-severity advisories surfaced by our automated scan.

Critical 0High 0Medium 0Low 0

Scanned 12d ago via OSV.dev · paperbanana (PyPI)

Adoption & maintenance

Factual signals from GitHub, npm, and our automated checks — not a rating.

GitHub stars2.4kStargazers on the source repository.
Last commit1mo agoMost recent push to the default branch.

How PaperBanana works

What PaperBanana MCP server does

PaperBanana MCP server turns textual descriptions of research methods into academic diagrams and produces statistical plots from structured inputs. It is aimed at AI-assisted research workflows where an agent needs to create a visual explanation of a model, system, or experiment rather than only return prose.

The implementation is an unofficial, community-maintained version based on the published PaperBanana paper. It is not affiliated with or endorsed by the paper’s authors or Google Research, so generated results and behavior may differ from the original system.

How it works

PaperBanana MCP server uses a multi-agent workflow with two main phases. A generation stage then creates the requested illustration or plot and can refine the result over multiple iterations.

The workflow supports automatic refinement and continuation of an earlier run with user feedback. Results include a final image as well as intermediate iterations and metadata when using the documented generation flow. PDF files can optionally provide methodological context, with page selection available when the PDF support is installed.

The project supports OpenAI models, Azure OpenAI or Foundry endpoints, Google Gemini, and Atlas Cloud. Provider selection and model behavior depend on the configured credentials and settings.

Setup and configuration

Install the Python package with pip install paperbanana. The documented prerequisites include Python 3.10 or newer and an API credential for either OpenAI-compatible access or Google Gemini. The repository also supports installation from source and provides a Docker workflow.

Set provider credentials through environment variables or a .env file. The documented variables include OPENAI_API_KEY and GOOGLE_API_KEY. Azure OpenAI or Foundry deployments can use OPENAI_BASE_URL; Gemini deployments can override the service URL and model settings. A setup wizard is available for Gemini configuration.

The package can run through its CLI, Python API, MCP server, or optional local Gradio Studio. Studio is installed separately and provides a browser interface for diagrams, plots, evaluation, batch jobs, run continuation, and browsing generated output folders.

Tools and capabilities

The PaperBanana MCP server supports agent-facing generation of academic diagrams and statistical plots from text. The surrounding project documents these related capabilities:

  • Generate methodology diagrams from an input text file and caption.
  • Produce statistical plots from CSV or JSON items in a batch manifest.
  • Optimize source text before generation.
  • Refine output automatically or continue a previous run with feedback.
  • Generate multiple diagrams or plots from YAML or JSON manifests.
  • Use selected PDF pages as additional methodology context.

The repository also exposes Claude Code skills for diagram generation, plot generation, and diagram evaluation, but those skills are separate from the general MCP interface.

Limitations and notes

Provider access requires the corresponding API credentials, and generated output depends on the selected provider and model. PDF support is optional and requires the PDF extra. Studio is also optional and requires its Gradio extra.

Review generated diagrams and plots before using them in research materials.

PaperBanana MCP server is an unofficial implementation of a research system.

Read the full README View source on GitHub

Reviews

No reviews yet — be the first to share how it worked for you.

Frequently Asked Questions about PaperBanana

PaperBanana is an MCP server that connects AI agents to model providers including OpenAI, Azure OpenAI or Foundry, Google Gemini, and Atlas Cloud. Its main capabilities are generating academic diagrams from text and creating statistical plots, with optional input optimization, iterative refinement, batch workflows, and PDF-based context.

Install it with `pip install paperbanana`, then configure a supported provider API key. The project also documents source installation and Docker usage.

PaperBanana supports `OPENAI_API_KEY` for OpenAI access and `GOOGLE_API_KEY` for Google Gemini. Azure OpenAI or Foundry can use an OpenAI-compatible base URL, while Gemini supports optional base URL and model overrides.

Yes, the repository documents Claude Code skills for generating diagrams, generating plots, and evaluating diagrams.

Yes, PaperBanana can run multiple statistical plots from a manifest, with CSV or JSON data specified per item. The project documents the `plot-batch` workflow and also exposes plot batch processing through PaperBanana Studio.

Related MCP servers

All in Data Visualization All alternatives
Add a badge to your READMEShow it’s listed — live SVG, Markdown or HTML.

AllMCPs Directory Badge

Full Badge Customizer

Showcase your server listing on GitHub or your project documentation. Embed this dynamic SVG badge to highlight official listing status and live engagement.

Badge Style:
Live Dynamic SVG PreviewPaperBanana AllMCPs Directory Badge
Markdown (GitHub README)
[![AllMCPs](https://allmcps.com/api/badge/paperbanana?style=directory)](https://allmcps.com/mcp/paperbanana)
HTML Embed
<a href="https://allmcps.com/mcp/paperbanana"><img src="https://allmcps.com/api/badge/paperbanana?style=directory" alt="PaperBanana on AllMCPs" /></a>