Lovable / AI-built app
Lovable AI Code Audit
A human review of generated code, Supabase access paths, authentication, Stripe integration and launch risks.
Explore the Lovable audit →Independent AI code audit / Production readiness
Your AI-built product works.
Know its production risks.
I provide independent, read-only technical audits of AI-built products. The review covers architecture, security, data access, payments and production readiness within an agreed scope. You receive evidence-backed priorities and recommendations for your team; implementation is a separate decision.
Business customers only · Read-only review · No implementation
Egidijus GirčysSoftware engineer & technical advisor
13+ yearsbuilding production software
Business-focusedtechnical risk in commercial context
AI-nativekeep the speed, add discipline
Choose the right review
Focused audits for the stack you use.
The general assessment covers an agreed product scope. These focused audits explain the specific checks, deliverables and purchase terms for Lovable and Supabase projects.
Supabase / Data security
Supabase Security Audit
A focused review of RLS, tenant isolation, Auth, Storage, Edge Functions and privileged access.
Explore the Supabase audit →01 / The assessment
AI code audits for vibe-coded products.
AI can get a product surprisingly far. The hard part is knowing whether the whole system is safe, maintainable and ready for real use. I review the product as a working business system—not just a repository—and give you a prioritized view of the risks, what should be fixed first, and what can wait.
Review areas within the agreed scope
- 01Architecture and code quality
- 02Security, privacy and access control
- 03Database and data integrity
- 04Tests and change safety
- 05Performance and scaling risks
- 06Deployment and observability
- 07Dependencies and operational risk
- 08AI workflow and engineering process
You receive
Technical Readiness Report
A prioritized, plain-English view of your product with evidence, business impact, and recommended actions.
CriticalImmediate attention
HighPriority before the next stage
MediumPlan deliberately
LowLower priority
+A prioritized set of recommended next steps
Illustrative report finding
See how a finding is explained.
This is synthetic example content, not an assessment of a customer system.
Example only · Not verified
Example: an access rule needs verification
- Illustrative observation
- A shared record access path may not enforce the intended tenant boundary. The exact behavior would need to be checked against the agreed code and existing tests.
- Potential impact
- A user could see data outside their intended account if the rule is missing across all access paths.
- Verification
- Illustrative only. No customer system or runtime behavior has been tested.
- Recommended next action
- Review the ownership rule and test access with two separate tenant identities.
02 / How it works
Clarity first. Then your next steps.
Understand
We discuss your product, next business milestone, current risks, and priorities.
Assess
I review the agreed code, configuration and existing evidence using read-only access or sanitized exports.
Prioritize
You get a direct report and a roadmap built around technical risk and business value.
Walk through
We walk through the findings and recommended actions. Your team decides, tests and implements any changes.
“The goal is clarity, not a rewrite.”
A pragmatic approach
Assessment first. Recommendations next.
The goal is not to manufacture problems or recommend a rewrite. I assess the agreed material and explain the risks and practical next steps. I do not modify your application; your team owns implementation and release decisions.
- Understand operational risks
- Prioritize material findings
- Prepare for diligence and scale
- Identify useful engineering guardrails
Egidijus GirčysSoftware engineer & technical advisor
About me
Experience matters more when code becomes cheap.
I’m based in Berlin, Germany and work with startups and product teams internationally. I’ve been building production software professionally since 2013, helping organizations move from early decisions to reliable systems, migrations, and healthier engineering processes.
I use AI extensively in my own work. The goal is not to replace fast AI-assisted development with slow “traditional” development. It is to add the architecture, verification, and operational discipline that lets you keep moving quickly without losing control.
Common questions
AI code audits with a defined scope.
Do you audit Lovable, Cursor, Replit and Supabase apps?
Yes. I provide business-only, read-only reviews of AI-built and AI-assisted products. The scope is agreed before purchase and may include code, architecture, authorization, data access, existing tests and deployment configuration. You receive a report and recommendations, not implementation.
Is this a certification or compliance audit?
No. This is a scoped, point-in-time advisory review for business customers. It is not a formal penetration test, legal opinion or compliance certification, and does not guarantee that the application is secure. You receive findings and recommendations; implementation is not included.
Do you work only in Berlin?
No. I’m based in Berlin, Germany and work remotely with startups and product teams internationally.
Ready when you are
Know what you are running
before the next big step.
Understand the material risks and the recommended next steps within the agreed review scope.
Book a 30-minute callFor business customers only. The agreed audit is advisory and read-only. Scope and terms