HQ Hermes / Control Plane HermesHQ

Docker-first / Managed Hermes runtimes

Operate multiple Hermes agents from one dark control surface.

HermesHQ keeps Hermes as the real execution engine, then adds the layer teams actually need in production: managed agents, tasks, schedules, Telegram, hierarchy-aware delegation, runtime profiles, and managed integrations.

Tasks Runtime ledger Comms Telegram Integrations
HermesHQ demo preview
Control plane Web UI + API
Runtime Real Hermes
Install Docker-first

Positioning

Hermes Agent vs HermesHQ

HermesHQ does not replace hermes-agent. It orchestrates it. Hermes alone is the execution engine; HermesHQ adds the multi-agent control plane around it.

Hermes Agent alone

Single runtime used directly

Local CLI/TUI usage, one local HERMES_HOME, direct prompt and plugin handling, and operator-managed local sessions.

HermesHQ

Control plane over managed runtimes

Separate workspaces per agent, RBAC, tasks, schedules, runtime ledger, comms, Telegram, runtime profiles, provider presets, secrets, and managed integrations.

Operational surfaces

Built for observation, dispatch, and follow-through.

The product surface is intentionally dark, information-dense, and operational. These are real screenshots from the running app.

HermesHQ dashboard
Dashboard and dependency canvas for agent fleet visibility.
HermesHQ agent detail
Agent detail with runtime settings, embedded TUI, conversation, ledger, and integrations.
HermesHQ comms
Comms surface with hierarchy-aware delegate routing and topology summaries.
HermesHQ integrations
Managed integrations with effective capabilities and per-agent configuration.

Runtime model

Same Hermes engine, managed as a real fleet.

HermesHQ uses Hermes as the real runtime, then standardizes how agents are configured, run, supervised, and traced.

Runtime profiles

Standard, technical, security

Profiles define effective capabilities. In the current phase, standard already blocks TUI and terminal/process usage, while technical and security profiles keep full operator workflows.

Managed integrations

Packages, not shell glue

Integrations are installed as packages, enabled per agent, and tested from the UI. This keeps API-based business integrations separate from core runtime concerns.

Hierarchy-aware comms

Delegate with real routing rules

Delegation is not cosmetic. Independent agents can delegate freely, escalation upward works, downward delegation within branch works, and cross-branch lateral delegation is blocked.

What the demo shows

One continuous workflow, not isolated screenshots.

The demo page uses a real player with controls and focuses on the surfaces that make HermesHQ feel like an operating console rather than a settings shell.

1

Create a new agent from the built-in runtime profile flow.

2

Open the agent detail surface and enter the real Hermes TUI.

3

Float and fullscreen the TUI to show the runtime as an operational surface, not just a text panel.

4

Review managed integrations and effective capabilities on a standard-profile agent.

5

End in Comms with hierarchy-aware delegate routing visible before dispatch.

Demo page

Video controls, dark layout, real product narrative

GitHub README is fine for documentation, but not ideal as a product presentation layer. This landing and the separate demo page carry the darker, more intentional HermesHQ feel.

Install

Docker-first, one-line setup.

HermesHQ installs and runs with Docker by default. The installer downloads the current branch tarball, writes or reuses .env, and runs docker compose up --build -d.

One-line install

Fast path

curl -fsSL https://raw.githubusercontent.com/jpalmae/hermeshq/main/install.sh | bash

If the server has multiple interfaces, pass HERMESHQ_HOST explicitly with the public IP or DNS name.

What you get

Persistent stack

Docker-managed PostgreSQL, persistent workspaces, generated bootstrap credentials when needed, and a backend/frontend pair ready to expose through your preferred ingress.