ArchGuard AI v1.2.0: I Added ChatOps and Custom Rules to my AI Architecture Reviewer

A few days ago, I published ArchGuard AI, a free, Zero-Trust Serverless GitHub Action designed to automatically review Pull Requests for architectural flaws (like Tight Coupling and Mass Assignment). The response from the community was amazing.

However, after watching developers use it, I noticed two major pain points:

1. Software architecture is subjective. What if the AI flags something as "Tight Coupling", but you actually had a good reason for doing it? You couldn't argue with the bot.
2. Every team has its own rules. The AI enforced generic clean architecture, but what if your CTO specifically wants everyone to use a custom EnvelopeResponse format?

Today, I’m excited to announce ArchGuard AI v1.2.0, which completely solves both of these problems with two killer features: ChatOps and Company-Specific Rules.

1. ChatOps: Argue with your AI Reviewer

Static code reviews are boring. Sometimes you just need to ask: "Why did you flag this?" or "Can you rewrite this using the Repository pattern?"

With v1.2.0, ArchGuard AI is no longer just a static linter. It’s an interactive bot living inside your Pull Request.

If you disagree with a review or want it to re-evaluate your latest commits, simply reply to the PR comment and tag the bot:

@archguard-ai please explain why this is a Mass Assignment vulnerability.
@archguard-ai I just pushed a fix, pls re-check.

The AI will read the context of the conversation, analyze the latest Git Diff, and reply to you in real-time. It feels exactly like pairing with a Senior Architect!

👉 Don't believe me? See a live demonstration in this real Pull Request: Live Demo PR #4 on Github

2. .archguardrules: Enforce Your Team's Discipline

ArchGuard AI is smart out of the box, but it can't read your CTO's mind. Until now.

You can now create a .archguardrules file in the root of your repository. This file acts as the absolute law for your repository.

For example, you can write plain text rules like:

1. All database queries must go through a Service layer. Never query Mongoose directly in a Controller.
2. Do not use standard HTTP responses. All APIs must return data wrapped in `{ success: true, data: ... }`.
3. Never use the Node.js `fs` module synchronously.

ArchGuard AI will automatically detect this file and aggressively enforce these rules on every Pull Request. It’s the ultimate way to maintain technical discipline across a large engineering team without having to constantly yell at junior developers.

3. End-to-End Daily Health Checks

Because ArchGuard AI runs on free Cloudflare Serverless infrastructure (Zero-Trust edge AI), we wanted to ensure 99.9% uptime. We have now implemented an open-source Daily Health Check CI.

Every night at midnight, our system automatically creates a dummy PR with intentionally bad code, waits for the AI to review it, and asserts that the architecture was correctly flagged. If the AI Gateway ever goes down, we know instantly.

The Promise: 100% Free, Forever

Unlike other AI tools that lure you in with a 14-day trial and then block you behind a paywall, ArchGuard AI is 100% Free for the community. We sponsor the entire Serverless Gateway infrastructure so you can use state-of-the-art LLMs without ever needing a credit card or an OpenAI API key.

Even better, it is built on a Zero-Trust Architecture. Your code hits the Cloudflare Worker RAM, gets processed, and is instantly destroyed. We do not store your source code, and we absolutely do not train models on it.

Just drop this into .github/workflows/archguard.yml:

name: ArchGuard AI Review
on: 
  pull_request:
    types: [opened, synchronize]
  issue_comment:
    types: [created] # Required for ChatOps
jobs:
  review:
    runs-on: ubuntu-latest
    permissions:
      pull-requests: write
      contents: read
      id-token: write # Required for Zero-Trust OIDC
    steps:
      - uses: archguard-labs/[email protected]

Check it out on the GitHub Marketplace and let me know how the ChatOps feels!