Your AI doesn't remember
what it built yesterday.
Find the contradictions in your AI-generated codebase.
Free · Runs locally · Zero data sent
NewSign up free and get 3 deep scans/month free — no card required→Spot the drift.
Same AI. Same project. Different sessions.
user_handler.ts
import { UserRepository } from '../repositories/user';
import { requireAuth } from '../middleware/auth';
export async function getUser(req: Request) {
requireAuth(req);
const user = await UserRepository.findById(id);
if (!user) throw new NotFoundError();
return user;
}✓ Repository pattern
✓ Auth middleware
✓ Typed error
order_handler.ts
import { db } from '../config/database';
export async function getOrder(req: Request) {
const id = req.params.id;
const rows = await db.query('SELECT * FROM orders...');
if (rows.length === 0) {
return { status: 404, error: 'not found' };
}
return rows[0];
}✗ Raw SQL — no repository
✗ No authentication
✗ Inline error object
Both pass every linter. VibeDrift catches this.
8 detectors · ~2s scan time · $0 free tier · zero bytes sent
What you get back.
A standalone HTML report. No login needed. Open it, share it, archive it.
VIBEDRIFT // my-saas-app
68/100Grade C
Architectural Consistency
9.6/25
Security Posture
25/25
Redundancy
7/20
Convention Adherence
12/15
Scaffolding Hygiene
15/15
ERRORArchitectural contradiction
orderService.ts:42 — raw SQL, 7 siblings use repo
ERRORMissing auth middleware
admin/reports.ts:8 — no requireAuth() guard
WARNSemantic duplicate (88%)
format.ts ↔ money.ts — same function, diff name
+ 14 more findings
Generated locally. Zero bytes leave your machine.
Want deeper analysis?
vibedrift . --deepAI finds semantic duplicates, intent mismatches, and validates every finding.
Three steps. No config.
01
Run.
No install.
No signup.
No config.
2 seconds.
No signup.
No config.
2 seconds.
02
Read.
Score: 68/100
Grade: C
5 categories.
Per-file breakdown.
Code evidence.
Per-file breakdown.
Code evidence.
03
Fix.
“Migrate orderService.ts to the repository pattern used by 7 other files.”
Every finding tells you
exactly what to change.
exactly what to change.
// Live context for your AI agent
Stop your AI from
drifting across sessions.
Every AI coding session starts cold. Claude Code, Cursor, Copilot — they see the file in front of them but not the eight sibling filesthat follow your project's repository pattern. So they happily write more drift.
VibeDrift writes a .vibedrift/ folder of peer-pattern context next to CLAUDE.md. Commit it. The next AI turn sees your dominant patterns, open drifts, and copy-paste fix prompts — no more guessing what the codebase actually looks like.
What gets written
.vibedrift/context.md— Project summary + dominant patterns + open drifts.vibedrift/fix-plan.md— Top 10 drift findings with copy-paste AI prompts.vibedrift/fix-prompts.md— Per-finding fix prompt — tackle one at a time.vibedrift/patterns.json— Machine-readable peer patterns for toolingSafe to commit. Every push is a fresh snapshot of your codebase's dominant patterns.
Two ways to use it
One-shot — run after a refactor:
Continuous — rescan on every edit:
Run alongside your Cursor / Claude Code session. Every file save refreshes
.vibedrift/ so the next AI turn sees up-to-date peer patterns. Zero network calls.Included on Free. Requires a one-time
vibedrift login.Pricing.
Free
Free
Try (no signup)
+Unlimited local scans
+Terminal preview (score + top fixes)
With free account
+Full HTML report + dashboard
+AI agent context (.vibedrift/)
+Watch mode for AI sessions
+AI fix-prompt synthesis
+3 deep scans / month
+Scan history + score badge
Pro
$15/mo
+Everything in Free, plus:
+50 deep scans / month
+$1/scan overage
+Support future development
Scale
$30/mo
+Everything in Pro, plus:
+100 deep scans / month
+$1/scan overage
+For codebases that scan often
Enterprise
Let's Talk
+Everything in Scale, plus:
+Self-host in your own cloud (BYOC)
+Per-seat team billing
+SSO / SAML
+GitHub App (auto PR)
+Custom drift rules + SLA
+Dedicated support
Every plan above starts from a free account. Unlimited local scans, full HTML report, AI agent context pipeline, watch mode, and 3 deep scans every month. No card required. Paid plans raise the deep-scan budget. Regulated industries, air-gapped environments, or strict data residency? Enterprise includes self-hosted deployment in your own cloud.
Stop guessing.
Start measuring.
Not ready to scan? Get notified of updates.