Here’s the full walkthrough. The Hermes OpenCode integration is how you run Hermes agent free forever: OpenCode’s free models become the brain, Hermes stays the hands, and a Kanban board turns them into an assembly line that builds while you do nothing.
We built a full website, a mini app and an analytics dashboard this way — free. Here’s the whole integration.
Last updated: August 2026.
Key takeaways
- Plug OpenCode’s free APIs and models into Hermes: OpenCode is the brain, Hermes is the hands.
- Run it through a Kanban board: Hermes triages tickets into subtasks and delegates every build to OpenCode.
- Failures retry themselves and Hermes verifies each artifact before moving on.
- The golden rule: Hermes never writes the HTML — orchestrator and builder stay separate.
- Get the full integration in the Agent OS via the AI Profit Boardroom.
How the Integration Works
Two free, open-source tools, one clean split. OpenCode is a superb, token-efficient coding agent with free models available — that’s the brain. Hermes is the always-on agent that orchestrates — the hands. Wire them together and Hermes hands every coding job to OpenCode, then checks the result.
Because both harnesses and the models are free, the whole system runs at effectively zero cost, 24/7.
The Kanban Assembly Line
The integration really sings on a Kanban board — an assembly line with five stations:
- The ticket — you drop a task on the board (“build a website about OpenClaw”).
- Triage — the Hermes dispatcher claims it and splits it into subtasks (research, architecture, design…).
- Delegation — each subtask goes to OpenCode, which writes every line.
- Auto-retry — failures go back on the board and retry until done, no re-prompting from you.
- Verification — Hermes checks the artifact before the ticket moves to done.
You come back to finished builds — the website, mini app and dashboard in the video were all produced exactly this way.
Old Way vs New Way
| Old way | With the integration | |
|---|---|---|
| Prompting | One agent, one prompt at a time | Tickets on a board |
| You are… | The message queue | Out of the loop |
| Failures | Manual re-prompts (tokens each time) | Retry themselves |
| Output | One build at a time | A pipeline of verified builds |
Why the Separation Matters
The trick that makes it reliable: Hermes’s profile is trained so it never writes the HTML itself — every build goes through OpenCode, and Hermes only orchestrates and verifies. Keeping the orchestrator and builder separate means each does the job it’s best at, and nothing gets half-built by the wrong agent.
For the deeper pieces, see my the delegation skill and the free models that power it — plus Hermes vs OpenCode for how the two compare head-on.
What You Need to Start
The shopping list is short: Hermes installed (free), OpenCode installed (free), and a Kanban board in your agent OS to run the assembly line through. Then set up the Hermes profile so builds route through OpenCode — or skip the setup entirely and tell Claude or Hermes to configure the system for you; that’s genuinely how I did it. Separate profiles per model make testing different free brains painless too.
Beyond Coding: Scheduled Workflows
Once the integration runs, you can schedule it. “Build an app every day based on my Obsidian memory and what would actually be useful” — and it does, analysing what you’re working on and shipping accordingly. The same pattern powers my lead-gen tool: generating leads, writing campaigns, managing inboxes — orchestrated by Hermes, built by the free brain.
The complete setup — Agent OS, Kanban, profiles and workflows — is in the AI Profit Boardroom.
New here? Start free with my AI Money Lab community (free AI course + 1,000+ AI agents), or grab a free strategy session.
FAQ
What is the Hermes OpenCode integration?
OpenCode’s free models plugged into Hermes as the brain, with Hermes orchestrating via a Kanban board and OpenCode writing every line — a free, 24/7 build pipeline.
Is it really free?
Yes — Hermes and OpenCode are free and open source, and OpenCode’s free models (plus alternatives like Ling 3.0 and OmniRoute) cost nothing. Rotate models if one rate-limits.
Why shouldn’t Hermes write the code itself?
Separation of duties: Hermes orchestrates and verifies, OpenCode builds. Each agent does what it’s best at, and the profile enforces it.
What did you actually build with it?
A full website, a mini app, an analytics dashboard and a Kanban tool — all through the board, all free, with retries handled automatically.
Do I have to babysit it?
No — that’s the point. Drop tickets on the board; the dispatcher triages, delegates, retries and verifies. You collect finished work.
The Bottom Line
The Hermes OpenCode integration is a free 24/7 build team: Hermes orchestrates, OpenCode builds, the board keeps it honest. Get it in the AI Profit Boardroom.
