Nous Research Releases Hermes Desktop: Native Cross-Platform Front End for Hermes Agent v0.15.2 with Streaming Tool

Nous Research released it Hermes Desktop in public preview. It is a native operating system for macOS, Windows, and Linux. It provides the open source Hermes Agent with a user interface. So far, users have run Hermes through the CLI and messaging gateways. The current build is Hermes Agent v0.15.2.

According to Nous Research documentation, the desktop also uses the same agent context. Share configuration, API keys, sessions, capabilities, and memory with the CLI and gateway. The desktop is an alternative to a single agent, not a fork.

What is Hermes Desktop

Hermes Agent is an autonomous AI agent. It is not a copy of code bound to the editor. It executes functions, calls tools, and maintains state at all times. An agent here refers to a model that plans, executes, and watches in a loop.

Hermes Desktop is a GUI on top of that agent core. It does not require a terminal to use. The window displays live streaming responses and live instrument activity. The right pane previews web pages, files, and tool results. It also includes a file browser, voice input and output, and a settings UI.

Sessions are shared across locations. A conversation started on the desktop resumes in the CLI or TUI. The reverse also works, because the situation is not repeated.

macOS and Windows offer direct installers. Linux installs from the terminal on any distribution. An installation script with --include-desktop the tag builds the application against the existing installation.

Closed learning loop

The Nous research team describes Hermes as having a closed learning loop. This is what separates it from a simple chat wrapper. After a complex task, the agent writes a skill that can be reused. Those skills then develop during later use.

Memory is persistent and selective, with occasional shifts to store information. Intermittent recall uses FTS5 session search with LLM summarization. User modeling uses the Honcho dialectic user model. In practice, longer usage means more stored context and reuse. Skills follow the agentkills.io open standard.

How to Connect, Schedules, and Sandboxes

Hermes runs across the message fields from one gate. Desktop includes Telegram, Discord, Slack, WhatsApp, Signal, Email, and CLI. You can start work in one place and continue in another.

Programming uses natural language for reports, backups, and summaries. These run unattended in the gateway in the built-in cron scheduler.

Delegation spawns individual subagents through their conversations with terminals. A subagent is a separate agent that handles a single task. Python RPC scripts collapse multi-step queues into zero-context-cost turns.

Execution is sandboxed. Desktop lists five backends: Local, Docker, SSH, Singularity, and Modal. Container consistency and namespace separation apply. Namespace isolation limits what a running process can see or touch.

Built-in tools include web search, automatic browser, visualization, image generation, text-to-speech, and multi-model reasoning. Hermes also connects external devices via MCP. MCP is the Model Context Protocol, a tool integration standard.

Hermes works with any provider, so API keys are optional. Nous Portal bundles them under one subscription instead. The portal tiers are free, premium, super, and ultra. Paid tiers include monthly credits and access to 300+ models. They also include using a built-in tool.

Tool Gateway supports several tools with one account. Web search uses Firecrawl and image rendering uses FAL. Text-to-speech uses OpenAI and the cloud browser uses Browser Usage.

Strengths and questions

Power:

• Native installers remove the installation requirement for most users
• Output streaming and previewing make tool calls easy to test
• Continuous memory and self-improvement skills reduce repetitive instructions
• Model-agnostic design avoids lock-in to a single provider
• The MIT license allows for evaluation, maintenance, and modification

Questions:

• The product is in public preview, so expect rough edges
• Automatic memory and planning raise questions of supervision and review
• The Linux desktop still installs through the terminal
• Broad capabilities mean a steep learning curve for beginners

Key Takeaways

• Nous Research has released Hermes Desktop in public preview, a native operating system for macOS, Windows, and Linux for its open source Hermes Agent.
• The GUI shares a single agent's context, configuration, API keys, sessions, capabilities, and memory with the CLI and gateway; sessions resume at all locations.
• It uses a no-terminal that features a streaming tool, side-by-side preview pane, file browser, voice I/O, and settings UI.
• Hermes is model agnostic and MIT licensed, works with Nous Portal, OpenRouter, OpenAI, and any compatible endpoint.
• The current build is Hermes Agent v0.15.2, with closed learning loop support, MCP tools support, and five sandbox backends.

Check it out Project here. Also, feel free to follow us Twitter and don't forget to join our 150k+ ML SubReddit and Subscribe to Our newspaper. Wait! are you on telegram? now you can join us on telegram too.

Need to work with us on developing your GitHub Repo OR Hug Face Page OR Product Release OR Webinar etc.? Connect with us