php-devops/doc/free-tier-compute.md
Snider a78ef46133 feat: add AI worker workflows (Gemini + Jules)
Distributed AI compute using contributor's free tiers:

Gemini 2.0 Flash (ai-worker.yml):
- 1500 req/day free from Google
- Code review, security scan, bug detection
- Contributor sets GEMINI_API_KEY in fork secrets

Jules/Copilot (jules-dispatch.yml):
- Triggered by @jules or /jules comments
- Creates PRs to fix issues automatically
- Uses contributor's Copilot allowance (free for OSS)

Documentation (doc/free-tier-compute.md):
- Setup guide for all free tiers
- Compute distribution model diagram
- Donor fleet instructions

Innocent strategy: Jules commits fixes to contributor's fork 😇

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-31 22:30:38 +00:00

3.6 KiB

Free Tier Compute Guide

This project uses distributed compute from contributors' free tier allowances. Here's how to maximize it.

GitHub Actions (Microsoft)

Free tier: 2,000 minutes/month for public repos

When you fork this repo, CI runs on YOUR allowance, not upstream's. This includes:

  • Full test suite
  • Linting and formatting
  • Security scanning
  • AI-powered triage

Gemini API (Google)

Free tier: 1,500 requests/day, 1M tokens/minute

Setup

  1. Get API key: https://aistudio.google.com/apikey
  2. Add to your fork's secrets:
    gh secret set GEMINI_API_KEY --repo YOUR_USER/YOUR_FORK
    

The ai-worker.yml workflow will use your Gemini allowance for:

  • Code review
  • Bug detection
  • Security analysis

Jules / Copilot (GitHub)

Free tier: Available with GitHub Copilot (free for OSS contributors)

Trigger Jules

Comment on any issue:

@jules fix the authentication bug in login.php
@jules add unit tests for the User model
@jules refactor to use async/await

Jules will:

  1. Analyze the issue
  2. Write code to fix it
  3. Create a PR on your fork
  4. Uses YOUR Copilot allowance

Get Copilot Free

CodeRabbit

Free tier: Unlimited for public repos

Already configured via .coderabbit.yaml. Reviews run automatically on PRs.

Compute Distribution Model

┌────────────────────────────────────────────┐
│           CONTRIBUTOR FORK                  │
│                                            │
│  ┌─────────────┐  ┌─────────────┐          │
│  │   GitHub    │  │   Gemini    │          │
│  │   Actions   │  │   2.0 Flash │          │
│  │   FREE      │  │   FREE      │          │
│  └─────────────┘  └─────────────┘          │
│                                            │
│  ┌─────────────┐  ┌─────────────┐          │
│  │   Jules/    │  │  CodeRabbit │          │
│  │   Copilot   │  │   FREE      │          │
│  │   FREE*     │  │             │          │
│  └─────────────┘  └─────────────┘          │
│                                            │
│  * With Copilot Pro (free for OSS)        │
└────────────────────────────────────────────┘
                    │
                    │ PR (already analyzed)
                    ▼
┌────────────────────────────────────────────┐
│           UPSTREAM REPO                     │
│                                            │
│  Just verify fork CI passed ✓              │
│  Cost: ~$0.001 per PR                      │
└────────────────────────────────────────────┘

For Donor Fleet Members

If you want to contribute compute without coding:

  1. Fork the repo
  2. Set up your API keys (Gemini, etc.)
  3. Keep Actions enabled
  4. Pick up agent:review tasks to verify others' work

Your free tier becomes part of the distributed compute network!