One Agent API
Expose agent work through a single API instead of wiring each harness separately. The page says you can integrate once and then choose the harness per task.
HarnessRouter
HarnessRouter provides one Agent API for plugging Codex, Claude Code, and Hermes into a product on self-hosted infrastructure or through a managed cloud option. Community Edition is open source and runs as a single Docker container that keeps keys, state, and files under your control.
Overview
HarnessRouter is an agent harness infrastructure product for plugging Codex, Claude Code, and Hermes into your product through one Agent API. The open-source Community Edition runs on your own infrastructure, while the page also points to a managed cloud option for teams that do not want to operate the backend themselves.
The Community Edition is packaged as a single Docker container that brings up the Gateway, Runner, and Console together. The site presents it as a way to keep your keys, state, and files under your control while using a public, versioned contract for harness implementations through the Unified Harness Protocol.
Core capabilities
Self-hosted container
Run the gateway, runner, and console together in one Docker container on infrastructure you control. The page notes there is no managed database, external vault, or cloud account required for Community Edition.
Local state and file persistence
Keep state, workspace files, and your provider keys on your own infrastructure. The page explicitly calls out SQLite state and a workspace volume in the local setup.
Production workflow primitives
Manage the full lifecycle around agent work, not just the initial request. Supported lifecycle pieces include sessions, streaming, files, artifacts, cancellation, and failure handling.
Unified Harness Protocol
Use the Unified Harness Protocol as a public, versioned contract. The page describes it as open, stable, reviewable, and independently testable, with OpenAPI 3.1 and JSON Schema 2020-12 support.
Starter Kit examples
Start from reference demos and production kits that show how sessions, streaming, files, and generated outputs fit into real product experiences. Examples include a cursor-style coding app, LumaCare, and Slides Kit.
Practical use cases
Self-hosted agent infrastructure
Use Community Edition when you need a self-hosted agent backend and want to keep provider keys, workspace files, and state on your own infrastructure.
Multi-harness integration
Use the single Agent API to route tasks to Codex, Claude Code, or Hermes without changing the rest of your product backend.
Agentic product workflows
Build experiences that need streaming work, resumable sessions, file uploads, artifacts, and cancellation rather than a one-shot model call.
Reference-driven product development
Start from the Starter Kit examples when you want a working product shape, such as a coding app, a care app, or a presentation workflow, instead of starting from an empty repo.
Managed deployment path
Choose the managed cloud option when you want the same API but do not want to operate agent infrastructure yourself.
Pros and Cons
Pros
- Supports multiple harnesses behind one API, including Codex, Claude Code, and Hermes.
- Open-source Community Edition is available under Apache 2.0.
- Runs as one Docker container with Gateway, Runner, and Console together.
- Keeps provider keys, state, and workspace files on infrastructure you control.
- Includes a managed cloud path if you prefer not to run the backend yourself.
Cons
- The source does not show broad integration coverage beyond the named harnesses and example product kits.
- Community Edition appears to require you to operate the container and its local state on your own infrastructure.
FAQ
How do I run HarnessRouter Community Edition?
Community Edition runs on infrastructure you control and is packaged as one Docker container with the Gateway, Runner, and Console starting together. The page shows a local quick start, and the UHP spec describes the public contract behind it.
Is there a free way to start?
The pricing page says you can build and integrate for free, and credits are consumed only when tasks run through your API key. A card is required to claim the 500 launch credits offered for a limited time.
What can I connect through the API?
The page highlights Codex, Claude Code, and Hermes as supported harnesses behind one Agent API. It also says you can select the harness for each task and switch without rebuilding your backend.
Can I change plans or add credits later?
Yes. The pricing page says upgrades take effect immediately with prorated billing, downgrades start at the next renewal, and top-ups can add prepaid usage credits without changing the plan.
When would I choose Cloud instead of Community Edition?
The open-source page positions Community Edition for self-hosted use and the managed cloud option for teams that want the same API without operating agent infrastructure.
Quick Facts
- Category
- Developer Tool
- Deployment
- Self-hosted open source or managed cloud
- License
- Apache 2.0 (Community Edition)
- Primary users
- Product teams building agent-enabled workflows
- Source domain
- harnessrouter.ai
- Pricing model
- Free to build and integrate; paid usage credits apply when tasks run
Alternativas ao HarnessRouter
CreateOS Sandbox
CreateOS Sandbox is an isolated compute environment for running code and agent workloads inside Firecracker micro-VMs. It is designed for workflows that need machine-level isolation, private networking between sandboxes, and programmatic control through SDK, CLI, or MCP.
ByteAsk
ByteAsk is a terminal-first AI coding agent for C and C++ that edits repositories and verifies changes with the real compiler, debugger, sanitizers, and tests before showing a diff. It offers a free tier plus paid plans, with editor connectors and zero-retention handling described in the source.
hob
hob is an independent workspace for coding agents that keeps agent sessions, terminals, history, and follow-up work organized around the tools and providers you already use. It is aimed at developers who want local control over routing, history, and workspace structure rather than a bundled model stack.
Manta AI
Manta AI is an autonomous web app testing tool for teams that want to map application behavior, catch regressions, and generate tests without writing scripts or maintaining selectors. It works from a URL and supports plain-English test flows, run results with screenshots, and scheduled or deployment-triggered checks.
SonOf
SonOf connects to your repo and PM tool, audits the codebase and surrounding product context, and turns approved work into shipped tickets with senior engineering review. It is aimed at founders and engineering leaders who need backlog help without hiring a full team immediately.
Ghost
Ghost é um assistente de IA para terminal, para conversar, gerar código e executar tarefas no prompt. Traz modelos gratuitos, funciona no Linux, macOS e Windows e é open source.