Nous Research has released Hermes Desktop in public preview. It is a native application for macOS, Windows and Linux. It gives a graphical interface to the open-source Hermes Agent. Until now, users ran Hermes through the CLI and messaging gateway. The current build is Hermes Agent v0.15.2.
According to Nous Research’s documentation, the desktop reuses the same agent core. It shares configuration, API keys, sessions, skills, and memory with the CLI and gateway. The desktop is another surface on top of an agent, not a fork.
What is Hermes Desktop?
Hermes Agent is an autonomous AI agent. This is not a coding copilot tied to an editor. It runs tasks, calls tools, and maintains the state of the entire session. An agent here means a model that plans, acts and observes in a loop.
Hermes Desktop is a GUI on top of the same agent core. No terminal is required to use it. The window shows streaming responses and live tool activity. The right-hand pane previews web pages, files, and tool output. It also includes a file browser, voice input and output, and a settings UI.
Sessions are shared across different surfaces. Conversation started in desktop resume in CLI or TUI. The reverse also works, because the state is not copied.
macOS and Windows provide direct installers. Linux installs from the terminal on any distribution. an install script with --include-desktop Flags builds the app against an existing install.
closed learning loop
The Nous research team describes Hermes as having a closed learning cycle. This is what makes it different from ordinary chat wrappers. After a complex task, the agent writes a reusable skill. Those skills automatically improve with subsequent use.
Memory is persistent and agent-curated, with periodic prompts to save knowledge. Cross-session recall uses FTS5 session search with LLM summary. User Modeling Honcho Dialectic runs through user modeling. In practice, longer usage means more preserved context and reuse. Skills follow Agentskills.io’s open standards.
How it connects, schedule and sandbox
Hermes runs from a gateway to a messaging platform. Telegram, Discord, Slack, WhatsApp, Signal, Email, and CLI are listed on desktop. You can start a task on one platform and continue on another.
Uses natural language for scheduling reports, backups, and briefings. These run unmonitored through the gateway on the built-in cron scheduler.
Delegation gives rise to separate sub-agents with their own interactions and terminals. A subagent is a separate employee who handles a job. Python RPC scripts collapse multi-step pipelines into a zero-context-cost solution.
Execution is sandboxed. Desktop lists five backends: local, docker, ssh, singularity, and modal.. It enforces container hardening and namespace isolation. Namespace isolation limits what a running process can see or touch.
Built-in tools include web search, browser automation, vision, image generation, text-to-speech, and multi-model reasoning. Hermes also connects external devices through the MCP. MCP is Model Context Protocol, a standard for tool integration.
Hermes works with any provider, so API keys are optional. Instead, Nous Portal bundles them under one subscription. Portal tiers are Free, Plus, Super, and Ultra. Paid tiers include monthly credits and access to 300+ models. These also include the use of built-in tools.
Tool gateway routes multiple tools through a single account. Web searching uses Firecrawl and image generation uses FAL. Text-to-speech uses OpenAI and cloud browser usage.
strength and questions
Strength:
- Native installer removes terminal requirement for most users
- Streaming output and preview tools make it easy to inspect calls
- Continuous memory and self-correction skills reduce repeated instructions
- Model-agnostic design avoids lock-in to a single provider
- MIT license allows audit, self-hosting, and modification
Question:
- The product is in public preview, so expect rough edges
- Autonomous memory and scheduling raise inspection and review questions
- Linux desktop still installs via terminal
- Wide capacity means steep learning curve for beginners
key takeaways
- Nous Research released Hermes Desktop, a native macOS, Windows, and Linux app for its open-source Hermes agent, in public preview.
- An agent shares core, configuration, API keys, sessions, skills, and memory with the GUI CLI and gateway; Sessions resume on all surfaces.
- It runs no-Terminal with streaming tool output, side-by-side preview pane, file browser, voice I/O, and Settings UI.
- Hermes is model-agnostic and MIT-licensed, working with Nous Portal, OpenRouter, OpenAI, or any compatible endpoint.
- The current build is Hermes Agent v0.15.2, which is supported by a closed learning loop, MCP tool support, and five sandbox backends.
check it out project here. Also, feel free to follow us Twitter And don’t forget to join us 150k+ ml subreddit and subscribe our newsletter. wait! Are you on Telegram? Now you can also connect with us on Telegram.
Do you need to partner with us to promote your GitHub repo or Hugging Face page or product release or webinar, etc? join us
Michael Sutter is a data science professional and holds a Master of Science in Data Science from the University of Padova. With a solid foundation in statistical analysis, machine learning, and data engineering, Michael excels in transforming complex datasets into actionable insights.
