Find the zero
in your code.
India gave mathematics the concept of zero. ArchSetu uses the same idea to find what's missing in your codebase: dead code, blind spots, and risk, before any of it costs you.
Live analysis preview · react/react
Not a demo
Run against real, working codebases, at real scale
These are live counts from ArchSetu's own database, the same query an engineer here would run to check the system, not a marketing estimate. Every number updates as more repositories get analyzed.
1,053
Real repositories analyzed
1,204
Completed analysis runs
24+
Languages, in production, not on a roadmap
768K
Functions parsed in a single run (torvalds/linux)
Static analysis across 15+ programming languages
What ArchSetu measures
Everything you need to understand your codebase
6 analysis features. No AI. No data stored. Works on any public GitHub repo instantly.
Blast Radius
Before you touch a function, know exactly what breaks. Trace call impact across 3 hops: upstream callers, downstream dependencies, affected files. No surprises in production.
Dead Code Detection
Find functions nobody calls. Clean them before they become liability. Grouped by file with safe-to-remove badges. Direct ROI: less code to maintain, test, and deploy.
Call Graph
See how every function in your codebase connects. Navigate visually with zoom, pan, and click-to-view. Works across 15+ languages without any setup.
Health Score
One number that tells you if your codebase is healthy or degrading. Weighted across complexity, dead code, test file ratio, duplication, and file size. Tracked over time.
File Explainer
Understand any file in the codebase instantly. No context-switching. Generated from structure, not AI. Deterministic, fast, always accurate.
Architecture Rules
Define rules. Enforce them in CI. Keep teams aligned automatically. Catch forbidden imports and complexity violations before they reach production.
Why ArchSetu, in an AI-written world
Code review keeps getting harder to trust. Ours does not guess.
AI coding assistants can write a working feature in minutes. They are far less reliable at telling you what that feature quietly broke, or what it left behind. ArchSetu is not another model summarizing your code. It is a deterministic analysis engine that reads your codebase the way a compiler does: structurally, completely, and the same way twice.
No guessing, only ground truth
Most tools ask a language model to summarize your code and hope it read it correctly. ArchSetu parses your actual syntax tree and traces your real call graph. Every number it shows you can be traced back to a specific line.
Built for how code gets written now
A growing share of the code in any repository was written by an AI assistant, not a person who remembers writing it. Nobody has full context on a codebase like that anymore. ArchSetu reconstructs that context on demand, from the code itself.
Same answer, every time
Ask a language model to review your codebase twice and you can get two different answers. ArchSetu runs the same static analysis pass whether it is Monday morning or the night before a release. Results are repeatable, not a matter of luck.
Your code never leaves your machine
Nothing is sent to a third-party model to be analyzed. Public repos are cloned into an isolated environment and deleted right after. Private repos and large repos run entirely inside your own GitHub Actions, under your own account.
How it works
From URL to full report in under 60 seconds
Paste any GitHub URL
Public or private repository. No browser extensions, no CLI to install. Just a URL.
Works with any github.com URL: repos, branches, forks.
ArchSetu analyzes the architecture
We clone the repo, run static analysis across all files, and build the call graph in under 60 seconds.
Isolated container. Clone deleted immediately after analysis.
Get a full shareable report
Shareable URL with 9 tabs: health score, call graph, dead code, blast radius, complexity, and more.
Share with your team. Embed in docs. Export to JSON.
What top open-source repos look like
Real analysis data from ArchSetu. Click any card for the full report.
Live data. Updated every hour. Click any card for the full call graph and dead code breakdown.
ArchSetu VS Code Extension
Bring the full power of ArchSetu directly into your editor. Get live complexity scores, dead code warnings, and call graph navigation without leaving VS Code.
- Inline complexity scores next to every function
- Call graph panel inside VS Code
- Dead code highlights as you type
Our Tools
Beyond the health score, deeper, sharper signals no other static analyzer computes. Every card below is a real, working tool, not a mockup.
Git-History Risk Scoring
Bus factor and ownership risk from a repo’s real commit history, not a static guess.
Try it now
Blast Radius Explorer
Pick a file, see exactly what breaks, the real reverse-dependency graph.
Try it now
Dead Code Finder
Functions with zero known callers, traced from the real call graph.
Try it now
Dependency Hygiene Checker
Packages imported in code but never declared anywhere.
Try it now
Security Risk Scanner
Hardcoded secrets, unsafe dynamic execution, string-built SQL. Sign in required.
Try it now
PR-Time Blast Radius Warnings
The Blast Radius Explorer above, but posted automatically as a comment on every pull request, before you merge, not after you ask.
In development
Config & Env-Var Dead Detection
Unused environment variables and orphaned config keys, dead code that lives outside your source files entirely.
In development
Repository Trust Score
One external-facing score combining health, security, ownership risk, and dependency hygiene - built for judging a repo before you depend on it.
In development
Architecture Drift Detector
Tracks your codebase against the layering rules you define, and flags exactly when and where it started drifting.
In development
Merge Risk Predictor
A conflict and breakage risk score for an open pull request, computed before anyone starts reviewing it.
In development
Technical Debt Timeline
Dead code, dependency staleness, and TODO density charted against commit history, so debt is a trend, not a guess.
In development
Code Ownership Map
A visual map of who actually owns which part of the codebase, reconstructed from real commit history, not a CODEOWNERS file nobody updated.
In development
Your code stays private
We clone your repository, run the analysis entirely on our servers, then immediately delete the clone. We store only the analysis results. Never your source code. No AI. No third-party code scanning. Every analysis runs in an isolated container.
Analyze your first repo free
Results in under 60 seconds. Sign up to save your history.
Get Started FreeNo credit card required. Works on any public GitHub repository.