Start a briefBrief
Build & Software

Code Audit & App Rescue in Singapore

If you own software you did not write and are now responsible for it, we review it and then finish it.

The brief

Who it’s for

For founders and SME owners whose app works in a demo but breaks with real users, holds real customer data, or has to pass a client's security review. Also for whoever has just inherited a codebase: a new owner, a new technical hire, or the person left holding it after a developer stopped replying. We are a poor fit if your app is locked inside a platform with no exportable source code, such as Base44, Bubble or Glide: there is nothing to review or take over, and we will tell you that in the first call rather than sell you a report.

That covers apps generated with Lovable, Replit, Bolt, v0, Base44 or Cursor, a codebase a developer left behind, and half-built projects an agency never delivered. You get a written review of what is actually there, ranked by severity, with a fixed quote for the work needed to make it safe to run. We are engineers based in Singapore, so the same people who write the findings ship the fixes. Penetration testing and managed SOC monitoring are outside this service. We agree the systems and review boundaries in writing before work begins.

Example engagement · for illustration

Understand the existing application before changing it.

An application is difficult to maintain or a previous developer has left. A bounded engineering review establishes what can be reproduced, what is failing and what evidence supports each repair recommendation.

A practical sequence

  1. 01

    Establish authorised repository access and reproduce the build in isolation.

  2. 02

    Trace selected failures, dependencies and critical user journeys.

  3. 03

    Prioritise fixes with a rollback plan and verification for each change.

An example of how the work could be scoped, not a client result. Your proposal sets the deliverables and acceptance criteria.

Agree before work starts

What completion looks like

  • Check 1

    Findings include reproducible evidence and a clear effect on users or operations.

  • Check 2

    A proposed repair has a regression check and a way to return to the preceding version.

Scope boundary

This is a software engineering review. Penetration testing, certification and unrestricted security assurance are outside this service.

6 parts

What’s included

  1. Access and ownership recovery

    Before any code review, we work out what you actually control — domain, repository, hosting, database, payment and email accounts — and what is still in someone else's name. This is the first thing to fix when a developer has gone quiet, because everything else depends on it.

  2. Source-code and architecture review

    A manual read of the codebase and how it is put together: authentication, authorisation, session handling, data model and where the business logic actually lives. This is the layer automated scanners cannot judge, and where inherited code usually hides its real problems.

  3. Configuration and access-control review

    On AI-built stacks most of the damage sits outside the repository. We review database row-level security policies, API keys and secrets exposed in the browser bundle, unauthenticated endpoints, webhook signature verification and storage permissions.

  4. A written report you can act on

    One page your investor or enterprise client can read, then findings ranked by severity with file and line references, a data-model note, what breaks as users grow, and a prioritised fix list. It is a point-in-time review of what we were given, not a certification that the software is free of defects.

  5. Remediation and finishing the build

    The part most reports leave out. Real authentication and roles, payments and webhooks that reconcile, database migrations and constraints, deploys you can roll back, and the missing features that stop it being a prototype. Quoted fixed-price against the findings.

  6. Handover you can actually use

    Accounts and secrets in your name, a runbook covering deploy, restore and rollback, and a walkthrough recorded so it survives your next developer. Your code, accounts and infrastructure stay in your name, and the fixes we make become yours once paid for.

Outcomes

What it changes

  • A straight answer on whether the software is worth saving, including when it is not
  • The people who write the findings are the people who ship the fixes — no report-and-run
  • Written findings you can hand to an investor, an enterprise client or your next developer
  • Your accounts, code and data back in your own name, not a former developer's
  • A Singapore-registered supplier on the contract, in your timezone, for software holding your customers' personal data
Tools and platforms
Next.js / ReactSupabase / PostgreSQLRow Level Security policiesNextAuth / ClerkStripe & webhook integrityPrisma / Alembic migrationsVercel / DockerGitHub / CI pipelines

Engagement and pricing

How it is sold

Repository size, reproducibility, dependency age and the number of critical journeys determine the review boundary.

A written scope identifies the deliverables, exclusions, dependencies and any recurring charges before work begins.

What to bring to the first conversation

Repository access through an approved channel, build instructions and examples of the most important failures.

Share a redacted example first. Access to accounts and sensitive records is arranged separately.

Explore the work and the decisions behind it.

Questions about

Code Audit & Rescue

Should I rebuild my AI-generated app from scratch, or fix what I have?

Usually fix. The data model and the user flows are normally the parts worth keeping, and they are the parts that took you longest to get right. The common defects (row-level security switched off, keys sitting in the browser bundle, no server-side validation, unverified payment webhooks, no migrations) are all repairable without starting again. Rebuild when one of three things is true: the data model cannot carry the features you need next, the code is tangled enough that no change is safe to make, or the platform will not give you your source code. That last one is decisive and worth checking first. If we look at your app and conclude you should rebuild, we will say so. It is a cheaper conversation for both of us than a rescue that was never going to work.

Do you do penetration testing or a security audit that proves the flaw is exploitable?

Penetration testing and managed SOC monitoring are outside this service. We review source code, configuration and access controls within an agreed scope, document findings and quote remediation. If your customer requires a penetration test, engage an appropriate specialist provider; our code review does not replace it.

How do I check whether my app's database is publicly exposed?

On Supabase and similar platforms, open your table editor and check whether Row Level Security is enabled on every table that holds user data, and whether each one has a policy that actually restricts rows to their owner. RLS switched off means anyone holding your anon key — which is published in your browser bundle by design and is trivially readable — can read the whole table. RLS switched on with a permissive policy is almost as bad and is the more common failure, because the dashboard shows it as protected. Then search your deployed frontend bundle for anything named service_role, secret or private key: a service-role key shipped to the browser bypasses every policy you have written. These are the two checks worth doing yourself today, before you hire anyone.

My developer has disappeared and I cannot reach them. What do I do first?

Recover access before you worry about the code. Establish who the registrant is on your domain, who owns the GitHub or GitLab organisation, whose account the hosting and database sit under, and who controls the payment gateway, the app store listings and the email records. In practice this is the step that decides whether you have a business, because a codebase you cannot deploy is worth very little. Domain registrars and hosting providers have documented account-recovery routes for a verified business owner, so start there and put the request in writing. Change nothing in production until you have a current backup you have tested. Only once access is settled is it worth assessing code quality.

How much does a code audit cost in Singapore?

In the Singapore market, a bounded review of a small-to-medium codebase typically runs in the low thousands, and larger architecture reviews with remediation planning run higher. For context, senior contract engineering in Singapore sits around S$1,200 to S$1,600 a day. The fix is almost always the larger number, and a serious supplier will tell you the range for that before you commit to the review. We give a fixed quote after a free scoping call rather than publish a rate, because a 2,000-file inherited codebase and a three-week-old Lovable app are not the same job. What we will not do is quote a review without a scope: an unbounded audit is how these engagements turn into open-ended consulting. Ours is bounded in writing to a stated codebase, a fixed window, and the findings that matter ranked by severity — explicitly not an exhaustive list of every defect.

Will you need access to my live database and my customers' personal data?

Not for the review, and we prefer not to have it. We work from the repository, a schema dump and redacted configuration, with real personal data excluded or replaced with synthetic records. That is enough to find the failures that matter, and it keeps your customers' data out of a third party's hands entirely. If a later remediation phase genuinely needs access to a live system, it happens under a signed data-processing agreement, on named accounts you create and time-box rather than a shared admin login, with the work done against a copy and no production changes outside an agreed window. Access is revoked in writing when the engagement closes. If a supplier asks for your production credentials before they have scoped anything, that is a reason to stop.

If my app has exposed customer data, is that a PDPA problem?

Potentially, yes. Singapore's PDPA Protection Obligation requires reasonable security arrangements over personal data in your control, and the failures the PDPC repeatedly cites in its published decisions — weak access controls, unpatched systems, poor oversight of whoever built the system for you — are the same defects that turn up in AI-generated and hastily inherited code. Responsibility stays with you as the organisation even when an outside developer wrote it, which is the part founders are most often surprised by. There are also notification duties once a breach meets the significant-harm or scale thresholds. We can map what personal data your app holds, where it flows and where it is exposed, and fix the technical side. That is an engineering assessment, not legal advice, and it is not a PDPA certification. If you need a formal legal position, speak to a lawyer or a qualified DPO.

Can this project receive grant support?

New applications moved to EDGE on 30 September 2026. Existing EDG, PSG and MRA submissions and projects continue under their original process. Check the exact activity, applicant conditions, supported costs, provider requirements and timing with Enterprise Singapore before relying on funding. GDI can prepare the technical scope and quotation; approval is the agency's decision.

All services
EXECUTIVE DISCOVERY · SINGAPORE SCOPING CALL

Tell GDI what you need built, fixed or run.

Start a project brief

Or WhatsApp +65 8714 1149 · or

A written reply, usually within one business day.