Hermes Agent Tools

Give your Hermes agent real-time capabilities

Flexura integrates seamlessly with Hermes Agent by Nous Research. Install once and your self-improving agent gains internet search, AI image generation, and every new capability Flexura releases — all through a single API key.

Works with Hermes skill system
Complements built-in learning loop
Real-time search + image generation
Model-agnostic — any LLM backend

What is Hermes Agent?

Hermes is an open-source, self-improving AI agent framework developed by Nous Research. Unlike traditional AI assistants that reset every session, Hermes features a built-in learning loop — when it completes complex tasks, it synthesizes the experience into reusable skill documents that it can recall and apply to similar problems in the future.

Hermes maintains a persistent model of your preferences and workflows using a layered memory system. It supports multiple LLM backends (OpenAI, Anthropic, OpenRouter, local models via Ollama), integrates with IDEs via Model Context Protocol (MCP), and can be deployed locally, on a VPS, or in serverless environments.

How Flexura enhances the Hermes learning loop

Hermes is designed to learn from every task it completes. When you ask Hermes to search for information using Flexura's API, the agent not only gets the result — it learns the pattern. The next time you need a similar search, Hermes already knows the optimal query structure, filtering parameters, and how to extract the most relevant information.

The same applies to image generation. After the first few uses, Hermes learns your preferred styles, resolutions, and prompt patterns. It builds procedural skill documents that make future image generation requests faster and more aligned with your expectations.

Real-time search for a persistent agent

Hermes runs as a persistent service — it remembers your previous conversations, project context, and decision history. But its LLM backend is still limited by its training data cutoff. Flexura's Search API fills this gap by giving Hermes access to real-time information from the entire internet.

Whether you deploy Hermes on your laptop, a cloud VPS, or a Docker container, the search capability works identically. Hermes can search for documentation, current events, API references, market data, research papers, or any topic — and incorporate the results into its persistent memory for future reference.

Multi-platform capabilities with Flexura

Hermes supports multi-platform interactions through Telegram, Discord, Slack, and IDE integrations. With Flexura installed, you can trigger search and image generation from any of these channels. Ask your Hermes bot on Telegram to research a topic, or tell it through VS Code to generate a UI mockup.

As Flexura adds new capabilities — text-to-speech, video generation, document reading — your Hermes agent gains them automatically. Combined with Hermes' learning loop, each new capability becomes more powerful over time as the agent learns your specific use patterns.

Frequently asked questions

Does Hermes remember how to use Flexura across sessions?

Yes. Hermes' built-in learning loop means it synthesizes successful Flexura API interactions into reusable skill documents. Over time, it becomes faster and more accurate at using search and image generation based on your specific patterns.

Which LLM backend works best with Flexura in Hermes?

Flexura works with any LLM that Hermes supports — OpenAI, Anthropic Claude, OpenRouter models, DeepSeek, or local models via Ollama and vLLM. The skill instructions are model-agnostic.

Can I use Flexura with Hermes deployed in Docker?

Yes. As long as the MEDIA_GATEWAY_API_KEY environment variable is set in your container, Hermes can use all Flexura capabilities. This works in Docker, Singularity, Modal, Daytona, or any deployment method Hermes supports.

Does Flexura integrate with Hermes' MCP support?

Flexura is installed as a skill file that Hermes reads directly. It complements Hermes' MCP integrations rather than competing with them — you can use MCP for IDE integrations and Flexura for search and generation capabilities.

Ready to add this capability to your agent?

One command. One API key. Works with Codex, Cursor, Windsurf, Claude, Aider, and Cline.