Skip to main content

AI Workflows: Copilot & PRs

Rigour is the "Engineering Handshake" between AI agents and your codebase. It ensures that regardless of which AI tool you use, the output adheres to your project's standards.

🤝 Workflow 1: GitHub Copilot (Advisor)​

GitHub Copilot is excellent at suggesting code, but it doesn't always know your project's specific engineering constraints (like complexity limits or forbidden dependencies).

In this workflow, Rigour acts as a Local Validator:

  1. Generate: Use GitHub Copilot to write a function or refactor a module.
  2. Audit: Before committing, run npx @rigour-labs/cli check.
  3. Refine: If Rigour identifies a violation (e.g., "Complexity too high"), use Copilot to fix it:
    • Prompt: "Copilot, Rigour says this function is too complex. Refactor it to reduce cyclomatic complexity by extracting the logic into a helper."
  4. Verify: Run rigour check again to ensure a PASS state.

🛡️ Workflow 2: Automated PR Reviews (Supervisor)​

Rigour can act as an Automated Technical Reviewer in your pull request pipeline. This ensures that only high-quality code reaches human review.

The Pull Request Pipeline:​

  1. Submit PR: A developer (human or AI) submits a pull request.
  2. CI Audit: Rigour runs automatically in GitHub Actions using --ci --json flags.
  3. Standardized Feedback:
    • PASS: The PR is marked as technically compliant. Humans can focus on business logic.
    • FAIL: Rigour blocks the merge and provides a rigour-fix-packet.json.
  4. Self-Healing: If an agent is handling the PR (like Claude Code), it can read the diagnostic packet and submit a fix commit automatically.

Why use Rigour in PRs?​

  • Consistent standards: Enforce the same rules across the entire team.
  • Save human time: Don't waste senior developers' time pointing out complexity issues or "missing tests" violations that Rigour can catch in milliseconds.
  • Safe delegation: Feel confident delegating large refactors to AI because Rigour acts as the safety rail.

🎯 Pro-Tip: The "Rigour-First" mindset​

When working with any AI, mention Rigour in your project instructions (e.g., in .cursorrules or .clinerules):

"Always verify changes using rigour_check. If violations occur, read the Fix Packet and resolve them before finalizing the task."