Project Workspace
3,400+ stars on GitHub

Your AI agent’s
command center.

Chat, memory, skills, terminal, and files — one interface for everything your agent does.

terminal
$curl -fsSL https://hermes-workspace.com/install.sh | bash

Setup

Running in three minutes.

One command installs Project Agent, clones the workspace, and wires the .env. Two more start it up.

Step 1

Install in one line

Detects Node 22+, Python 3.11+, pnpm — installs what's missing, installs hermes-agent from PyPI, clones the workspace, and configures .env. Re-runnable.

terminal
$curl -fsSL https://hermes-workspace.com/install.sh | bash

Step 2

Start the gateway and the UI

Two daemons. Project Agent on :8642 handles models, memory, and skills. The workspace on :3000 is your interface.

terminal
$hermes gateway run# terminal 1 · :8642
$cd ~/hermes-workspace && pnpm dev# terminal 2 · :3000

Or run pnpm start:all to launch both at once. First run prompts for hermes setup to pick a provider.

Compatible withAnthropic, OpenAI, OpenRouter, Atomic Chat, Ollama, vLLM, LM Studio, and any OpenAI-compatible /v1/chat/completions endpoint. Pick during hermes setup.

Screens

Section by section.

Real product screens across the core workflow.

Screen 01

Chat

Multi-model conversations with real-time tool activity.

Chat screenshot

Screen 02

Conductor

Mission orchestrator — spawn parallel agents, watch them work.

Conductor screenshot

Screen 03

Dashboard

At-a-glance metrics across sessions, messages, tools, tokens.

Dashboard screenshot

Screen 04

Memory

Browse and search what your agent remembers.

Memory screenshot

Screen 05

Terminal

Browser-native pty inside the workspace.

Terminal screenshot

Screen 06

Settings

Themes, providers, accents — 8 paired light/dark themes.

Settings screenshot

Features

Built for agents that do real work.

One interface for the parts of your agent that usually scatter across terminals, tabs, and logs.


6

core surfaces

100+

skills

PWA

mobile parity

MIT

open source


01

Multi-model chat

Claude, GPT, Gemini, Codex, and local models in one conversation flow. Switch mid-thread without losing context.

02

Memory & skills

Persistent memory plus a deep catalog of 100+ skills your agent can pull from. Browse and edit live.

03

Integrated terminal

Run commands where your agent lives — pty in the browser, full color, no leaving the workspace.

04

Real-time tool cards

Watch the agent think and act with live tool states streaming into the UI. Approve, abort, inspect.

05

Mobile-first PWA

Full feature parity on desktop, tablet, or phone. Install to home screen, run offline, push notifications.

06

Conductor + Operations

Mission orchestrator with parallel sub-agents, live worker grid, and an Operations console for managing every running agent.

Install