AI-powered repo audit and product-readiness engine

AI turned words into code. Harmonia turns code into product readiness.

Point Harmonia at a repository and get a readiness score 0–100 and your top 10 fixes in pain order — what's dead, what's faked, what leaks, what blocks deploy. Layer 1 runs locally; your code never leaves your machine.

Check public repo — free Install local CLI Polish version

Public preview uses the GitHub API · private audits run locally · no code upload.

Architecture 0 source modules Security 0 files scanned Tests 0 passing Deployment 0 pipelines Product coherence 0 / 100 · this repo
Live check

Check your repo's readiness. Free, one scan.

Paste a public GitHub repo. We read its structure in your browser via the public GitHub API — nothing is uploaded — and show a readiness preview with up to 3 fixes. Secrets, dead code and the full list? That's the CLI.

Public repos only · one scan per repo · code never leaves your browser. Private repos & the full report (secrets, dead code, semantic sense QC) — install the CLI.

Free live preview

Public repo structure only, straight in the browser. Instant signal — not the full audit.

Sees: public GitHub structure.

CLI · GitHub Action

The full offline audit on your machine: score 0–100 + every fix in pain order.

Your code never leaves the machine.

Pro Sense QC

Server-side semantic sense QC: the README's promise vs. real delivery. Licensed API.

Sees only report facts + a README excerpt — never your code. Engine never bundled.
What it actually is

You wrote the code.
Is it a product yet?

You write code. You think you're done — and yet nothing really runs. Endpoints aren't wired. The build is half-finished. Config that worked on your laptop dies in production.

And there's no help anywhere — just a folder of generated files that looks finished.

That's the gap Harmonia Repo Ready closes: point it at a repo and it tells you, in order of pain, exactly what stands between you and a real product.

The new problem

Generated code is not product readiness.

Harmonia Repo Ready — proof & investor kit
Repo Ready report on a real GitHub repo — repo data fogged, report readable Investor deck — page 01 Investor deck — page 02 Investor deck — page 03 Investor deck — page 04 Investor deck — page 05 Investor deck — page 06 Investor deck — page 07 Investor deck — page 08 Investor deck — page 09 Investor deck — page 10 Investor deck — page 11 Investor deck — page 12 Investor deck — page 13 Investor deck — page 14 Investor deck — page 15 Investor deck — page 16 Investor deck — page 17

Authentic screens & investor deck — looping. Self-audited 90/100 · CI green · Repo Ready v0.3.0 — real numbers, not mockups.

AI can generate code quickly, but it does not always preserve the coherence of the full product. It can add endpoints without UI, create components nobody uses, produce tests that check easy things instead of important flows, or generate configuration that works locally and fails in production.

Is the frontend connected to the backend?
Is authentication actually safe?
Can the project deploy?
Do the tests protect the critical flows?
Does the documentation match the code?
Is this a product, or a folder full of generated files?
What Harmonia Repo Ready does

It audits AI-built repositories for real product readiness.

Product Readiness Score

Determines whether the repository is ready as a product, not just as code.

Security Readiness

Reviews auth, roles, secrets, API exposure and production configuration.

Architecture Map

Shows how the system is actually structured across frontend, backend and infrastructure.

Broken Flow Detection

Finds disconnected UI, unused endpoints, dead code and duplicated logic.

Deployment Review

Checks whether the project can survive outside the developer's machine.

Fix Plan

Turns findings into a prioritised plan instead of another vague audit document.

● Readiness in action

The same offline audit. Three very different verdicts.

Run it across the repos on your disk and you get the truth, in order of pain — from “mockup, build the foundations first” to “ready to show”. Layer 1 is fully local: your code never leaves your machine.

20/100
Mockup — build the foundations first
AI-built MVP, never cleaned up
  • ⛔ .env committed to the repository
  • ⛔ Looks like a secret in scripts/runtime.ts
  • ⚠️ No CI · no tests · no .env.example
62/100
Prototype — not investor-ready yet
Works on the laptop, not handoff-ready
  • ⚠️ Only 2 test files — a fig leaf
  • ⚠️ Reads env vars, but no .env.example
  • ℹ️ No deployment signal (Docker/Procfile)
92/100
Ready to show
Clean enough to hand off / pitch
  • ✅ Tests + CI in place, no secrets in code
  • ℹ️ A stray TODO or two to triage
  • ℹ️ Missing LICENSE — clarify legal status
● Repair in action

Watch the same repo climb 20 → 92.

Close the fixes in order of pain and readiness moves with you. A credit clears only when the finding verifiably disappears with no new criticals — a failed repair costs nothing.

20
Start: .env + secret committed, no tests/CI
→
48
Removed secrets from code & history, revoked keys
→
78
Added CI + smoke tests on the critical paths
→
92
.env.example + LICENSE → ready to show

Free today: the honest verdict + ordered fix list — hand it to your agent or team. Coming next: “Fix with Harmonia” — verified repair, credit-based.

View sample report Audit your repo Free · local · offline. One honest number, plus what to fix first.
Where Harmonia stands

We don't compete with code generators. We start where they stop.

they generate · Harmonia verifies · one symbiosis

Different layer, not a rival — a symbiosis. We admire the tools that got us here: Cursor, Copilot, Claude Code, Codex, v0, Bolt, Lovable, Replit Agent. They are the elder siblings of this era, and Harmonia begins exactly where they finish. Their output is our input — we run after them, on your machine, and tell you the truth before you ship. No overlap, no fight: the checkpoint the AI-build era was missing.

* What runs where. Free / Layer 1 is offline, on your machine — heuristics for structure, tests, CI, secret patterns, dead weight and deploy signals. The semantic verdict — README-vs-delivery and broken-flow reasoning — is Layer 2 (Pro / Harmonia API). The web live check covers public repos and shows the top findings; the full list runs in the CLI.

From AI-built chaos to launch-ready software

Harmonia does not compete with tools that generate code. It verifies whether generated code has become a real product.

01 / Language

A founder, developer or agent describes a product, task or change.

02 / Code

AI generates components, endpoints, tests, configuration and documentation.

03 / Risk

The repository grows faster than its coherence, safety and deployment readiness.

04 / Harmonia

Harmonia audits the repository and verifies whether it is ready to ship, show or scale.

Pricing

Simple, honest pricing.

Free forever for the local audit. Local scans stay generous; AI reports and verified repairs are metered. Flip any card for the full feature list and limits.

Free
£0
Curious devs · testers
  • 1 repo
  • 5 scans / month
  • 1 sample report

Included

  • Layer 1 audit — fully offline
  • Readiness 0–100 + top fixes
  • CLI + GitHub Action + MCP
  • Public-repo live check

Limits

  • Layer 1 only — no semantic sense QC
  • 1 watermarked PDF · no history
  • Live check: 3 findings, 1 scan/repo
Founder deal
Founder Pro
£19/mo
Solo founders · freelancers · £29 after launch
  • 3 repos
  • 50 scans / month
  • 10 AI reports

Included

  • Everything in Free
  • Full fix list + semantic sense QC
  • Markdown / PDF export (20/mo)
  • History & deltas · private repos
  • 25 fix credits / month

Limits

  • 1 user
  • AI reports & credits metered
Studio
£99/mo
Freelancers · small studios
  • 20 repos
  • 400 scans / month
  • Client-ready PDF reports

Included

  • Everything in Pro
  • 80 AI reports · 200 exports /mo
  • Client handover reports
  • Audit history · 3 users
  • 150 fix credits / month

Limits

  • Fair-use caps apply
Team
£259/mo
£219/mo billed annually · save 15%
Software houses · teams
  • 75 repos
  • 1,500 scans / month
  • Team access · priority

Included

  • Everything in Studio
  • 250 AI reports · 500 exports /mo
  • Org-wide thresholds
  • Shared reports · priority queue
  • 10 users · 400 fix credits

Limits

  • Fair-use caps apply
Agency / White-label
£499/mo

High-volume audits, white-label client reports, custom limits and seats — readiness at scale for software houses.

One-off Repo Audit
£239

No subscription — a full audit of one repo plus a shareable PDF. For a single decision: ship, hand over, or show an investor.

Fix with Harmonia
from £49 · credits from £29

Verified repairs, charged only when the finding verifiably disappears — a failed repair costs £0.

Lock the £19 founder rate

Founder Pro: £19/month until launch — locked forever while your subscription stays active.

● Fix with Harmonia · coming soon

Don't just find it. Fix it — and pay only when it's actually fixed.

The free check finds what's wrong. Fix with Harmonia closes it: a full-repo audit, a priced repair plan, and verified fixes you approve before paying a penny.

1
Full repo audit
The deep scan — secrets, dead code, README-vs-delivery — not just structure. You get a full report.
2
Repair estimate
Every finding priced in credits (≈ £). You see the bill before anything runs.
3
Approve & pay
Yes or no — your call. Secure checkout by Stripe. Pay only for fixes you greenlight.
4
Verified report
A credit clears only after the finding verifiably disappears with no new criticals.
Repair estimate · example
37 credits ≈ £81–116
You pay only for fixes that land. A credit is charged after the finding verifiably disappears with no new criticals — a failed repair costs £0.
🔒 Secure checkout · opening at launch
StripePayPalRevolut
Pre-book Pro — lock founder pricing Full live check — full repo (secrets, dead code, full report) ships with Pro.

From ancient harmony to AI-built software, Harmonia brings order, verification and readiness to the new world of AI-generated products.

Start repo audit Polish version