AgentIndex icon
AgentIndex
ToolsMCP ServersHOTCategoriesTrendingNewBest ListsCompareGuides
Submit Tool
ToolsMCP ServersHOTCategoriesTrendingNewBest ListsCompareGuides
Home/
LLM Infra/
harnessrouter
harnessrouter logo

harnessrouter

Active·★ 3.2k·Apache-2.0·Updated 2026-10-22
★ Essential★ Trending

HarnessRouter calls itself the unified interface for agent harnesses. The README says it turns existing harnesses such as Codex and Claude Code into plug-and-play agent backends, so a product can run tasks, retrieve results and switch harnesses through one API. It implements the Unified Harness Protocol (UHP) and provides an OpenAI Responses-compatible API, handling persistent sessions, streaming progress, files, artifacts, cancellation and structured failures.

The README includes a benchmark chart comparing eight harness and model configurations on the same task and states that the lowest-cost and fastest configurations vary by task; the methodology is on the project's website, so treat the figures as the project's own. This repository is the self-hosted Community Edition, with your own provider keys on infrastructure you control, and a managed Cloud offering is linked from the site.

Quick start is one Docker command that binds to 127.0.0.1:3000 and keeps data in a named volume. It needs Docker, about 4 GB of disk and a provider API key; no HarnessRouter account is required. The first launch installs the enabled harness CLIs, and then you sign in to the console, add a provider under Bring Your Own Key and run a first task. The README says the entrypoint needs root to manage per-session users, so you should not add --user, and it warns that the console starts with a default username and password that you must change in Profile before exposing the instance. It is Apache-2.0 licensed and the last push was 2026-10-15. This review is based on the README and repository data; we have not run it ourselves.

This write-up is based on the project's README and public repository data. We have not hand-tested the tool. Data as of 2026-10-16.

License
Apache-2.0
Latest release
v0.34.1 · 2026-10-15
Last commit
2026-10-16
Contributors
27
Forks
324
Open issues and PRs
14
First published
2026-08-09

Source: GitHub, npm and PyPI public APIs.

#Agent Orchestration#LLM Infrastructure#API Gateway#Unified Protocol#Developer Tooling#Containerization#Backend Integration#Agent Harness
$ Install
$ docker run -d --name harnessrouter -p 127.0.0.1:3000:3000 -v harnessrouter:/data harnessrouter/harnessrouter
↗ Visit site★ GitHub
01

Features

01Unified API for multiple agent harnesses (Codex, Claude Code, etc.)
02OpenAI Responses-compatible API for seamless integration
03Handles persistent sessions, streaming progress, files, and cancellation
04Self-hostable Community Edition with Docker for local control
05Supports custom harness configuration with tools and skills
02

Why choose it

+One OpenAI Responses-compatible API in front of harnesses such as Codex and Claude Code
+Handles persistent sessions, streaming progress, files, artifacts, cancellation and structured failures
+Self-hosted Community Edition with your own keys; no HarnessRouter account needed
+One Docker command to start, bound to localhost by default; Apache-2.0 licensed
03

Trade-offs

!The console ships with a default username and password; change it in Profile and keep the instance on localhost until you do
!The container runs as root internally and the README says not to add --user, so isolate it as you would any agent runner
!First launch downloads harness CLIs and needs about 4 GB of disk plus your own provider key
!The cost and latency benchmark is the project's own, and the managed Cloud edition is a separate commercial offering
04

Maintenance check

ActiveChecked: 2026-10-22
Last commit
2026-10-21
Commits, last 90 days
100+
Releases, last 90 days
30
Latest release
v0.35.14-rc.1 (2026-10-22)
Open issues
26
Recent issues with at least one comment
2/12
Commits per week, last 12 weeks
Commits per week, last 12 weeks: 0, 92, 59, 76, 58, 77, 50, 51, 74, 54, 23, 43

Oldest week on the left; the latest four weeks are orange. Source: GitHub commit activity.

Status rule: archived on GitHub = Archived; last commit within 90 days = Active; 91 to 365 days = Slowing; older = Stale. All figures come from the GitHub API on the date shown.

Most discussed open issues

  • #451 Console: a produced file's card sometimes does not render though the turn's record has the file4 comments · 2026-10-12
  • #25 Should HarnessRouter support persistent harness runtimes instead of spawning a CLI process per turn?4 comments · 2026-08-23
  • #479 UEP: First-class input requests — `input_required` event, `awaiting_input` status, and `POST /v1/responses/{id}/input`2 comments · 2026-10-17
  • #476 UEP: `requires_input` — a task can end by asking the user, and the answer resumes it2 comments · 2026-10-16
  • #382 The DeepSeek Harness driver keeps its own copy of the relay, without the runner relay's repairs2 comments · 2026-10-04

Issue titles are quoted from GitHub. Open the link to check whether a problem still applies to the version you use.

05

Compatibility

Docker
Deployment
Verified via docs
OpenAI Responses API
API Standard
Verified via docs
Unified Harness Protocol (UHP)
Protocol
Verified via docs
Codex
Agent Harness
Verified via docs
Claude Code
Agent Harness
Verified via docs
06

Quick start

1
$ docker run -d --name harnessrouter -p 127.0.0.1:3000:3000 -v harnessrouter:/data harnessrouter/harnessrouter
07

Use cases

↳Integrating diverse agent backends into a single product API
↳Building agentic features for knowledge work, such as content creation and data analysis
↳Optimizing cost and latency by switching between different harnesses/models
↳Developing agent-powered starter kits for slides, sheets, dashboards, or videos
↳Self-hosting agent infrastructure to maintain control over data and credentials
08

Alternatives

⚡Top VS:
VSconduit★235→VSos-moda★120→VSCherry Studio★52.9k→VScua★30.7k→
conduit logo
conduit★ 235
Local MCP gateway that cuts your agent's tool-token overhead ~90%. Every MCP server's tools collapse to 3 meta-tools the agent searches on demand. Measured: 97% less per request, same results. Keys in your OS keychain, no cloud.
VSCompare→
os-moda logo
os-moda★ 120
Your server fixes itself at 3am. You sleep. AI-native operating system built on NixOS.
VSCompare→
Cherry Studio logo
Cherry Studio★ 52.9k
A powerful desktop client for multiple LLMs. Supports local and cloud models.
VSCompare→
cua logo
cua★ 30.7k
Open-source infrastructure that gives AI agents computers they can use: sandboxed desktops and VMs, drivers for operating native apps, SDKs and benchmarks for computer-use agents.
VSCompare→
E2B logo
E2B★ 14.5k
Open-source infrastructure for running AI-generated code in secure, isolated cloud sandboxes, with SDKs for code execution and desktop computer use.
VSCompare→
bifrost logo
bifrost★ 9.0k
Fastest LLM gateway (50x faster than LiteLLM) with adaptive load balancer, cluster mode, guardrails, 1000+ models support & <100 µs overhead at 5k RPS.
VSCompare→
agent-harness-generator logo
agent-harness-generator★ 713
🛠️ The meta-harness for AI agents — scaffold your own focused, branded agent harness with its own npx CLI, MCP server, memory, learning loop, and witness-signed releases. Works with Claude Code, Codex, pi.dev, Hermes, OpenClaw, and RVM (hardware-isolated sandbox).
VSCompare→
runanywhere-sdks logo
runanywhere-sdks★ 10.3k
Production ready toolkit to run AI locally
VSCompare→
See all alternatives →

Comments

Log in to leave a comment

No comments yet. Be the first!

On this page
01Features02Why choose it03Trade-offs04Maintenance check05Compatibility06Quick start07Use cases08Alternatives
HarnessRouter /
harnessrouter
↗

Apache-2.0 self-hosted Community Edition that turns Codex, Claude Code and other agent harnesses into one OpenAI Responses-compatible API

3.2k324Apache-2.0
Last commit1d ago
StatusActive
CategoryLLM Infra
Contributors27
Latest releasev0.34.1
Links
Documentation↗Discussion↗Issues↗Releases↗
Cloud Infrastructure
Ad

Deploy on DigitalOcean

Host your MCP servers, Docker containers, and AI agents with $200 free credit over 60 days.

Claim $200 Free Credit
© 2026 AgentIndex.app|Built by a 10-year iOS Developer.
Best by use caseAboutContactAdvertisePrivacy PolicyTerms of UseQYSGitHubBuy me a coffee ☕

Browse by Category

MCP ServersCode AssistantWorkflow AutomationRAG / Knowledge BaseMulti-AgentBrowser AutomationLLM InfraDev ToolingObservability

Not affiliated with Anthropic, OpenAI or Microsoft.