Automate GitHub Issues
With Your AI

Works with Claude Code, Cursor, Ollama, and LM Studio. Use one or mix and match. Switch between them anytime. Arbiter handles the Git workflow. Your AI handles the code.

Arbiter Dashboard
Dashboard Screenshot Placeholder
Take a screenshot of your Arbiter dashboard and save it as
dashboard-screenshot.png

24/7

Autonomous Operation

100%

Local-First & Secure

Issues Solved

Works With These AI Workers

Use one or use them all. Mix and match across repos. Switch anytime. Paid or free—you decide.

Claude Code

Claude Code

Official Anthropic CLI

Recommended
Cursor

Cursor Agent

Cursor's agent system

Fast & Powerful
Ollama

Ollama

Local open-source models

100% Free
LM Studio

LM Studio

Local models with GUI

Easy Setup

More Coming Soon

Codex and more

In Development

Arbiter handles: Issue scanning, Git operations, test running, PR creation, error handling

Your AI handles: Reading code, writing solutions, understanding context

Complete Workflow Orchestration

Arbiter automates everything around your AI—so it can focus on writing great code

Complete Workflow Automation

Arbiter automates the entire Git workflow—scanning issues, managing workspaces, running tests, creating PRs. Your AI focuses solely on writing code. Controlled, consistent, predictable.

Multi-AI Support

Supports Claude Code, Cursor, Ollama, and LM Studio. Use one, or mix and match across repos. Switch between them anytime. Arbiter handles orchestration, your AI handles coding.

Local-First & Secure

Runs entirely on your machine. Your code never leaves your control. Full data privacy guaranteed.

Validation & Testing

Runs your test suites and validation commands before committing. No broken builds, ever.

Smart Continuations

If work is interrupted, Arbiter automatically resumes where it left off. Never loses progress.

Real-Time Monitoring

Beautiful web dashboard shows live progress, logs, and history. Jump in anytime to see what's happening.

Draft PRs

Opens pull requests as drafts by default. Review before merging—you stay in control.

Self-Learning System

Learns from each PR it creates. Builds knowledge of your codebase patterns and conventions over time.

PR Iteration

Comment on a draft PR with changes, and Arbiter automatically implements them. Iterate until it's perfect—no manual back-and-forth.

Secondary AI Review

Have a second AI review the first AI's work before creating a PR. Catch issues early with dual validation.

Webhook Integration

Connect Arbiter to Slack, Discord, or custom endpoints. Get notified when issues are solved, PRs are created, or errors occur.

Advanced Analytics

Track every run with detailed logs and metrics. See what worked, what failed, and optimize your workflow over time.

Set It and Forget It

Four simple steps to autonomous issue solving

1

Label Your Issues

Add a label like agent-ok to GitHub issues you want Arbiter to handle. That's it—Arbiter does the rest.

2

Arbiter Finds & Prioritizes

Arbiter automatically scans your repos for labeled issues. Smart prioritization ensures important work gets done first.

3

Arbiter Orchestrates, AI Codes

Arbiter sets up the workspace and hands off to your AI. Your AI reads the issue, understands the codebase, and writes the solution. Arbiter then runs tests, validates the changes, commits the code, and creates the PR. Consistent, controlled, predictable.

# Arbiter orchestrating the workflow...
Arbiter: Analyzing issue #42: "Add dark mode toggle"
AI: Reading codebase context...
AI: Implementing solution...
Arbiter: Running tests... passed
Arbiter: Creating pull request...
4

Review & Merge

Wake up to draft PRs ready for review. Arbiter did the heavy lifting—you just approve and merge.

Pay What You Think Is Fair

Honest pricing with no hidden fees. 7-day free trial, no credit card required.

Monthly

$5/mo
Billed monthly
Pay $3-$50/month. You choose!
  • 1 machine
  • Unlimited repos
  • Unlimited issues
  • Claude Code, Ollama, or LM Studio
  • Web dashboard
  • Priority support
  • 7-day free trial
Start Free Trial

Student discount: 50% off with valid student ID. Contact us

Our Philosophy

Why We Don't Auto-Renew

Most SaaS companies make it hard to cancel. We do the opposite.

You Choose to Renew

Your Arbiter license doesn't auto-renew. When it's time to renew (monthly or annually), you'll receive a reminder. If Arbiter is valuable to you, renew with a single click. If not, no hard feelings—your license simply expires.

Why?

We believe people should pay what they think is fair for tools they genuinely find valuable. When you choose to renew each period, you're actively deciding Arbiter is worth supporting. That keeps us honest and focused on building something you actually want to use.

💜

Support What You Use

Your renewals directly support continued development, bug fixes, and new features.

🎯

No Dark Patterns

No hidden charges. No sneaky renewals. Just honest, transparent pricing.

Your Choice

Love Arbiter? Renew instantly. Need a break? No problem. You're always in control.

Pro tip: Annual plans save you 15% and lock in your price for the year. Monthly plans give you flexibility to reassess each month.

Common Questions

Everything you need to know about Arbiter

Is my code secure? Does Arbiter send data to external servers? +
Arbiter is completely local-first. Your code never leaves your machine. The AI workers (Claude Code, Ollama, or LM Studio) run on your hardware. The only external connection is to GitHub's API for reading issues and creating PRs—which you control.
What if Arbiter makes a mistake? +
Arbiter opens all PRs as drafts by default. You review every change before it goes live. Plus, Arbiter runs your test suite before committing, so broken code is caught early. You're always in control.
How does Arbiter work with AI models? +
Arbiter is an orchestration layer—it handles all the Git/GitHub automation while your AI handles the coding. You bring your own AI: Claude Code, Cursor Agent, Ollama (local models like Llama 3.1, Qwen, Mistral), or LM Studio. Arbiter manages issue scanning, workspace setup, test execution, and PR creation. Your AI just writes code. This separation keeps outcomes predictable and controlled.
What does Arbiter cost vs. AI costs? +
Arbiter is $3-50/month (you choose what's fair). Your AI costs are separate and depend on what you already use: Claude Code or Cursor Agent have API costs (typically $0.10-0.50 per issue solved), while Ollama and LM Studio run locally for free (just electricity). Arbiter doesn't add AI costs—you bring your own.
What types of issues can Arbiter solve? +
Arbiter excels at bug fixes, feature additions, refactoring, documentation updates, and test writing. Best for well-defined issues with clear acceptance criteria. Complex architectural changes may need human oversight.
Does Arbiter work with private repos? +
Absolutely! Arbiter works with both public and private repos. Just provide a GitHub personal access token with the appropriate permissions.
Can I run Arbiter on a schedule? +
Yes! Arbiter has a built-in scheduler. Set it to run every N minutes, or run it manually whenever you want. Perfect for overnight processing.
What's the refund policy? +
We offer a 7-day free trial—no credit card required. If you're not satisfied within the first 30 days of purchase, we'll refund you in full, no questions asked.

Ready to Automate
Your GitHub Issues?

Start your free 7-day trial. No credit card required.

Start Free Trial →