Archify — Architecture Quality Review for VS Code
Know your codebase's real strengths and weaknesses — before they hit production.

Archify is an AI-powered architecture reviewer that scores your codebase across 72 quality attributes in 10 categories — scalability, security, reliability, maintainability, and more. Get an instant architecture health report with actionable insights, right inside VS Code.
Works with any LLM you already use — GitHub Copilot, Claude, Amazon Q, or any BYOK provider. No extra API keys needed.
How It Works
- Open any project in VS Code
- Click the Archify icon in the sidebar
- Scan — Quick Scan (~30s) or Full Review (~2 min)
- Explore your interactive architecture report
What You'll See
Architecture Dashboard
Your project's health at a glance — overall score with letter grade (A+ to F), score history across scans, and top findings prioritized by severity. One click to scan, review, or export.
Interactive Report
A 4-tab deep dive into your architecture:
- Summary — Overall score with radar chart, AI-written executive summary, project stats, and quick wins you can ship today
- Categories — 10 expandable cards. Drill into any of the 72 attributes to see exactly what was found, what's missing, and how to fix it
- Findings — Every issue sorted by severity, with effort estimates and business impact. Fix, discuss, or dismiss — your call
- Export — Download as JSON or CSV. Cloud sync coming soon
Chat Integration
Talk to Archify in GitHub Copilot Chat:
@archify /review — Full architecture review
@archify /score — Quick quality score
@archify /explain — Plain-English architecture summary
@archify /deps file — Dependency map for any file
Also available as Copilot tool references: #archify_review, #archify_scan, #archify_deps
What Gets Evaluated
Archify analyzes 72 attributes across 10 categories — going far deeper than syntax or linting:
| Category |
What It Checks |
| Performance & Scale |
Scalability, throughput, latency, caching, concurrency, elasticity |
| Reliability & Resilience |
Fault tolerance, graceful degradation, idempotency, redundancy |
| Security & Compliance |
Authentication, authorization, secret management, audit trails |
| Maintainability & Ops |
Testability, observability, deployability, debuggability |
| User Experience |
Accessibility, responsiveness, internationalization |
| Integration & Portability |
API compatibility, modularity, extensibility |
| Efficiency |
Resource usage, cost efficiency, reusability |
| Business Continuity |
Disaster recovery, backup strategies |
| Code Architecture |
Coupling, cohesion, separation of concerns, simplicity |
| DevOps & Evolution |
Rollback support, reproducibility, evolvability |
Attributes that don't apply to your project are automatically marked N/A.
Each attribute receives a score, grade, verified findings, gaps identified, and actionable suggestions.
Scoring
| Grade |
Score |
Meaning |
| A |
90–100 |
Excellent — well-architected |
| B |
80–89 |
Good — minor improvements possible |
| C |
70–79 |
Average — notable gaps |
| D |
60–69 |
Below average — needs attention |
| F |
0–59 |
Critical — significant risks |
Works With Any LLM
Archify uses the VS Code Language Model API, so it works with whatever LLM you already have installed:
- GitHub Copilot — works out of the box
- Claude for VS Code — Anthropic's Claude models
- Amazon Q — AWS AI assistant
- Any BYOK extension — bring your own key
No separate API keys or accounts required for core functionality.
Security
- API keys stored in OS keychain via VS Code SecretStorage
- CSP-enforced webviews with cryptographic nonces
- Workspace-scoped file access only
- Formula injection protection on CSV exports
Archify Cloud (Coming Soon)
Shared team dashboards, quality trends over time, branch comparisons, and benchmarking. The extension works fully offline — cloud features are optional.
Requirements
- VS Code 1.93.0+
- Any LLM extension (GitHub Copilot, Claude, Amazon Q, or BYOK)
Feedback
Have a bug report or feature request? Reach out at dineshksaini.com or connect on LinkedIn.
License
MIT — Dinesh Kumar Saini