Skip to main content
Factor analyzes your pull requests automatically once installed on your GitHub account. In a few minutes, you’ll have Factor running on your repositories and surfacing risks on every new PR.

​
⚡ Step 1: Install Factor

Install Factor on your GitHub account. During installation, you’ll select:
  • Which organizations to connect
  • Which repositories Factor can access
Once installed, Factor gains read access to pull requests and repository context required for analysis. 👉 No code changes required.

​
⚙️ Step 2: Configure repositories

After installation, choose the repositories you want Factor to monitor. You can:
  • Enable Factor on all repositories
  • Or select specific services (recommended for first setup)
Factor will only analyze repositories you explicitly enable.

​
🔄 Step 3: Start your first analysis

That’s it. Once configured, Factor automatically runs on:
  • New pull requests
  • New commits in open pull requests
  • Updated PR descriptions or changes
No manual triggers required.

​
🧠 What happens next

When a PR is opened, Factor will:
  1. Analyze code changes
  2. Map system dependencies
  3. Simulate failure scenarios
  4. Generate a structured risk report
You’ll see:
  • Risk verdict (SAFE / REVIEW_REQUIRED / BLOCKER)
  • Potential failure scenarios
  • Missing tests or validations
  • Key assumptions in the code

​
📌 Example workflow

​
⚠️ What you don’t need to do

  • No code instrumentation
  • No CI modifications
  • No manual triggers
  • No configuration per PR
Factor works in the background once installed.

​
🧪 First run recommendation

We recommend starting with:
  • One active backend service
  • A medium-complexity repository
  • A team that ships frequently
This helps you immediately see meaningful analysis signals.

​
🚀 Next step

Once Factor is running:
  • 👉 Open your first pull request
  • 👉 Review Factor’s analysis
  • 👉 Start identifying missing tests and hidden risks

​
🧠 Notes

Factor is most effective on:
  • backend services
  • distributed systems
  • business logic-heavy codebases
Less signal is expected on:
  • simple UI changes
  • formatting-only commits