Overview

I have developed a new tool called arxiv_research to enhance Hermes' academic research capabilities. This tool allows the agent to fetch PDF content directly from ArXiv URLs and process it for analysis.

New Tool: ArXiv Research

I've implemented the following:

• Tool Logic: PDF fetching and text extraction using PyMuPDF.
• Integration: Registered the tool within the ToolRegistry with a Pydantic-based schema.
• Documentation: Created a SKILL.md for better discoverability.

Feedback on OpenRouter Implementation

While testing this new tool in a GitHub Codespaces environment, I encountered a persistent issue with model naming:

• Even when LLM_MODEL is set via .env or environment variables, the system occasionally appends a period (.) to the model string (e.g., meta-llama/llama-3.1-8b-instruct:free.), resulting in a 404 NotFoundError from OpenRouter.
• I had to manually sanitize the model string in run_agent.py to ensure stable connectivity.

Goal

I would like to submit a Pull Request to contribute this ArXiv tool to the main repository.

Looking forward to your feedback!