Real-time frontend feedback while you code with AI.
Vipr is an at-rest, multi-surface analyzer for front end development. It watches your repo while you (or your agent) code, and it flags anti-patterns and complexity the moment they show up. The CLI is live on npm today. The desktop app and VSCode extension are in private alpha with a small founding cohort.
Why Vipr
Built for
TypeScript-native depth
25+ analyses tuned for React hooks, Next.js boundaries, and TS-specific patterns. Not Java rules retrofitted for JS.
Your code stays yours
Analysis runs locally. No cloud upload, no telemetry on source, no account required to run a scan.
Speaks to your AI stack
An embedded MCP server lets Claude, Cursor, Antigravity, Kilo Code, and Windsurf query your repo metrics while they edit.
Three pillars
Built for code you ship every day.
Depth
25+ analyses across your codebase
Cyclomatic, cognitive, Halstead, maintainability, dead code, React anti-patterns, Next.js boundary leaks. One tool, one pass.
Learn moreContinuity
Quality tracked sprint over sprint
Per-commit snapshots, velocity trends with inflection-point detection, and churn-weighted hotspot rankings.
Learn moreAgency
Your AI agent stops guessing
An embedded MCP server gives Claude, Cursor, Antigravity, Kilo Code, and Windsurf direct access to real metrics from your codebase.
Learn moreFile detail, in context
Open a file. See why it scored that way.
Health score, the metrics behind it, threshold context, and AI guidance. One screen. Read the number, see what moved it, decide what to fix.
A value referenced inside useEffect is not listed in its dependency array, so the effect closes over the value from a previous render and silently uses stale data until the component re-mounts.
useEffect(() => fetch(url, { filters }), [url]);
The effect captures `filters` from the first render but never lists it in the dep array, so users see filtered results from the previous query until the next prop change forces a re-mount.
Overview
Build the dashboard you actually open every morning
Drag, drop, and pin the widgets that match how your team works. Score, hotspots, anti-patterns, budgets. Arrange them once and they stay put.
Per-file metrics
Every file scored, every metric tracked
Composite health for each file, broken down by complexity, churn, and maintainability. Trend pills show what moved since the last commit.
Composite Health Score
Churn × complexity
Four quadrants, one refactor queue
Concerning, Risky, Active, Stable. Files plot themselves by churn and complexity, so the work order is obvious before the meeting starts.
What it tracks
One engine, twenty-five analyses, zero cloud.
- Analyses
- 25+ Cyclomatic, cognitive, Halstead, churn, anti-patterns, dead code, and the rest.
- First analysis
- ~2 min 300 files, scored end-to-end, before your coffee finishes brewing.
- Bytes uploaded
- 0 Local engine, local DB, local MCP. Your code never leaves the machine.
Hotspots
The files that change most are the files that hurt most
A treemap sized by impact and colored by risk. The dense, dark squares are where churn meets complexity. Those are the refactors that pay back fastest.
- Risk Score
- 87.4
- Risk Level
- High
- Complexity
- 64
- Dependency Impact
- 28
- Hotspot Issues
- 6
- Lines of Code
- 412
Dependencies
Trace the blast radius of any file
Walk upstream to see what pulls a module in. Walk downstream to see what breaks if it changes. Cycles get flagged the moment they form.
Day zero
Open a repo, get a verdict in minutes
A first-pass readout of overall health, top hotspots, and architectural shape. Useful on legacy code you inherited and on greenfield work that just shipped.
Architecture Review
Identify architectural anti-patterns, circular dependencies, and high coupling that may impact maintainability.
Differentiation
Not a linter. Not a formatter. Your sidekick.
Linters catch style and syntax. AI review guesses at intent. Vipr sees the whole picture.
| Capability | Vipr | ESLint / Biome | AI code review |
|---|---|---|---|
| Understands React (hooks, boundaries, props) | ✓ | ✕ | ~ |
| Tracks quality as you code | ✓ | ✕ | ✕ |
| Can run 100% offline | ✓ | ✓ | ✕ |
| Works with AI agents via MCP | ✓ | ✕ | ✕ |
| Flags framework anti-patterns, not style | ✓ | ~ | ~ |
| One-time license, not per-seat or subscription-based | ✓ | ✓ | ✕ |
Surfaces
Same engine, four ways to use it.
The CLI ships today. The IDE extension, Desktop app, and MCP server are in closed alpha. Click through to preview each surface.
CLI
The same engine you can install today. Terminal output, JSON for CI, Markdown for PRs.
Private Alpha IDE
VSCode, Cursor, Antigravity, Kilo Code, Windsurf, VSCodium. File metrics, hover cards, and AI guidance next to the code.
Private Alpha Desktop
A local-first dashboard for the whole repo. History, hotspots, dependency graph, MCP server.
Private Alpha MCP Server
Lets your AI agent query real metrics from your code instead of guessing. Runs on localhost.
Pricing
Free CLI forever. One-time payment for the rest.
No subscription. No per-seat fees. One purchase covers three devices and every update inside that major version. Powered by Polar.sh.
CLI
Free forever
Analyze any JS/TS repo from your terminal. No account, no ceiling.
VSCode Extension Pro
$49 once
Inline scores, hover cards, CodeLens, dashboard webview.
Desktop Pro
$99 once
The full local dashboard. History, hotspots, MCP server.
Want lifetime upgrades? Believer at $249.99 one-time, every future major version included.
(Lifetime means all future major-version releases tied to your license key.)Founding Testers · 100 seats
We're picking 100 people to break this with us.
The deal is simple. Install Vipr, run it on a real repo each week, and tell us what's wrong. In exchange, you lock in 50% off the Bundle when v1 ships — the desktop app and VSCode extension together. We read every issue.
Cohort closes when we hit 100 testers or when the build is ready — whichever comes first.
You've already applied. We'll be in touch.