outsourc-e/hermes-workspace

⭐ 3,781  ·  JavaScript  ·  GitHub Repo

Native web workspace for Hermes Agent — chat, terminal, memory, skills, inspector.

agent-ui ai-workspace hackathon hermes-agent nous-research react typescript

1-Sentence Summary

Native web control plane transforming Hermes Agent from CLI into a visual multi-agent workspace.

🔥 Key Capabilities & USP

• Swarm Mode with Kanban TaskBoard — Deploy persistent, tmux-backed Hermes Agent workers (builder, reviewer, ops, triage, QA) and orchestrate autonomous workflows using a visual Kanban board. Solves the pain of managing multiple CLI agent sessions and tracking complex, multi-step tasks.
• Multi-Agent Control Plane — Monitor and manage unlimited Hermes Agents from a single dashboard with live panels showing queue depth, history, and usage meters. Eliminates the need to juggle separate terminal windows for each agent.
• Memory & Skills Management — Browse, search, and edit agent memory with a live markdown editor; manage a catalog of 2,000+ skills with origin badges and marketplace integration. Provides a visual alternative to opaque, file-based memory and skill management.
• Full Workspace Tooling — Real-time SSE streaming chat with tool call rendering, a Monaco-powered file browser, and a cross-platform PTY terminal. Unifies chat, code editing, and shell access into one coherent interface.
• Operations Dashboard — Multi-agent profile presets (Sage, Trader, Builder, Scribe, Ops), a cost ledger, model mix overview, and automatic 'Needs setup' detection. Turns agent management from guesswork into a data-driven operation.

USP: Unlike simple chat wrappers, Hermes Workspace provides a complete, native web control plane with persistent swarm-mode workers, role-based dispatch, and a Kanban task board for autonomous agent operations — all without forking the upstream Hermes Agent.

Technical Architecture

Component	Technology / Approach
Frontend	React + TypeScript, PWA-capable
Backend Connection	SSE streaming to Hermes Agent gateway (port 8642) and dashboard API (port 9119)
Security	Auth middleware on every route, CSP headers, path-traversal guard, fail-closed remote bind
Deployment	Docker Compose, one-line install script, or attach-to-existing Hermes Agent
Worker Management	Persistent tmux-backed processes for swarm mode
Theming	Hermes, Nous, Bronze, Slate, Mono (light + dark variants)
Capability Gates	Clean placeholders for features requiring upstream plugins (e.g., Conductor)

Quick Start Guide

One-line install (recommended):

curl -fsSL https://raw.githubusercontent.com/outsourc-e/hermes-workspace/main/install.sh | bash

Start services (two terminals):

# Terminal 1: Start Hermes Agent gateway
hermes gateway run

# Terminal 2: Start the workspace
cd ~/hermes-workspace && pnpm dev

Attach to an existing Hermes Agent:

git clone https://github.com/outsourc-e/hermes-workspace.git
cd hermes-workspace
pnpm install
cp .env.example .env
echo 'HERMES_API_URL=http://127.0.0.1:8642' >> .env
echo 'HERMES_DASHBOARD_URL=http://127.0.0.1:9119' >> .env
pnpm dev

Verify services are running:

curl http://127.0.0.1:8642/health
curl http://127.0.0.1:9119/api/status

Pros, Cons & Use Cases

Pros

• Zero-fork design — Works with vanilla Hermes Agent via Nous's official installer, ensuring compatibility with upstream updates.
• Flexible deployment — Choose between Docker Compose, one-line install, or attaching to an existing agent setup.
• Persistent context — Tmux-backed workers maintain state across tasks, enabling long-running autonomous workflows.
• Security-first — Auth middleware, CSP headers, and path-traversal guards are built-in, not afterthoughts.
• Byte-verified review gate — Protects release branches from corrupted or malicious changes.

Cons

• Conductor feature is placeholder-only — Requires an upstream dashboard plugin not yet in vanilla Hermes Agent, limiting multi-agent orchestration capabilities.
• Strict dependency requirements — Requires Node.js 22+ and specific Hermes Agent services (gateway + dashboard) running.
• Complex remote setup — Tailscale/LAN configuration for API URLs requires careful manual setup for remote access.

Who should NOT use this?

• Users without Hermes Agent installed — This is a control plane for Hermes Agent, not a standalone application. It provides zero value without the upstream agent running.
• Users on Node.js < 22 — The project explicitly requires Node.js 22+, so older environments will fail.
• Users wanting a simple chat interface — If you just need a basic AI chat UI, this is overkill. The workspace is designed for multi-agent orchestration and autonomous workflows.
• Production-critical deployments — The project is fresh from a hackathon (3.7k stars, recent creation). While promising, it lacks the maturity and testing for mission-critical enterprise use.

Ideal Use Cases

• AI researchers running multiple Hermes Agent instances for experiments, needing to compare memory, skills, and outputs visually.
• Home server power users who want a unified dashboard to manage autonomous agent workflows (e.g., code review, content generation, data analysis).
• Developers building agentic systems who need to orchestrate role-based swarms (builder, reviewer, QA) with a visual task board.
• Teams exploring multi-agent architectures who want a ready-made control plane without building from scratch.

Community & Activity

With 3,781 stars and recent updates (last updated May 2026), Hermes Workspace has clearly struck a chord with the AI agent community. The project's rapid star growth suggests strong interest in visual agent management tools, especially those that don't require forking upstream projects. The hackathon origins and active development signal a project with momentum — one to watch closely as the Conductor feature and upstream plugin support mature. If you're in the Hermes Agent ecosystem, this is the workspace you've been waiting for.