📝 Integration Proposal: CAJAL — Local Scientific Paper Model

[Agnuxo1]

📝 Integration Proposal: CAJAL — Local Scientific Paper Model for ChatBox

What is CAJAL?

CAJAL is not a general-purpose chat model. It is a specialized scientific paper generation tool — local, 2GB, producing LaTeX-formatted academic output.

Part of P2PCLAW

Agent in P2PCLAW — 14-agent decentralized research network.

Why ChatBox?

ChatBox users already switch between models. Adding CAJAL gives them a "Research Assistant" persona — not chat, but paper generation:

• Dedicated model card: "CAJAL — Scientific Writer"
• System prompt optimized for IMRaD structure
• Export to LaTeX/PDF via P2PCLAW

Proposed Integration

1. Add CAJAL to model registry — GGUF via Ollama/llama.cpp backend
2. "Research Mode" UI — distinct from chat, focused on paper drafting
3. Export integration — draft → P2PCLAW peer-review pipeline

Links

• 🤗 HF: https://huggingface.co/Agnuxo/CAJAL-4B-P2PCLAW
• 📁 GH: https://github.com/Agnuxo1/CAJAL
• 🌐 P2PCLAW: https://p2pclaw.com

— Francisco (@Agnuxo1), P2PCLAW

[mashrum]

Can it be done via MCP ?

[Agnuxo1]

Hi Maxim,

Great question! Yes, CAJAL can absolutely be integrated via MCP (Model Context Protocol).

Here is how it would work:

• MCP Server: Exposes CAJAL's paper generation pipeline as tools (generate_section, verify_citations, tribunal_score)
• MCP Client (Chatbox): Users trigger paper generation through Chatbox's interface, which routes to the local CAJAL server
• Local-first: The MCP server runs on the user's machine (Ollama backend), so zero data leaves the device

We have not built the MCP wrapper yet, but it is on the roadmap for Q2. If Chatbox is planning MCP support, we would love to collaborate on making CAJAL one of the first scientific writing tools available through it.

Would you be open to a joint PR, or should we prototype it independently and ping you when ready?

— Francisco (Agnuxo1)