DEV Community

Cover image for Technical Due Diligence Checklist for a Software Stack
Muhammad Ali Mughal
Muhammad Ali Mughal

Posted on

Technical Due Diligence Checklist for a Software Stack

A technical due diligence review is not a code beauty contest. When an investor or acquirer sends engineers to look at your software, they are answering three money questions: do you actually own what you are selling, will it keep running after the deal closes, and how much will it cost to fix what is broken. Founders who prepare for those questions close faster and negotiate from a stronger position. Founders who do not tend to discover their problems at the same moment the buyer does.

I run Loop Origin, a software development company that builds and maintains Laravel and Vue/Nuxt platforms for growing businesses. I wrote this with help from AI tools and checked every command and claim myself. This is engineering guidance, not legal advice.

What is a technical due diligence checklist?

Direct answer: A technical due diligence checklist is the list of evidence a buyer or investor reviews to judge your software: code and IP ownership, open-source licenses, architecture, security and secrets, infrastructure cost, data protection, delivery process and key-person risk. Preparing it in advance turns surprise findings into a planned, priced fix list.

How reviewers actually grade findings

Every finding lands in one of three buckets, and knowing them changes how you prepare:

Severity What it means for the deal Typical examples
Deal-breaker Deal pauses until fixed or restructured Code written by a contractor with no IP assignment; customer data breach not disclosed; core module under a license that forbids your use
Price adjuster Fixed via escrow, holdback or lower valuation Framework several major versions behind; no tested backups; one engineer who understands billing
100-day plan Noted, fixed after close Thin test coverage; missing runbooks; untidy CI

Your goal is not zero findings. It is no deal-breakers, and price adjusters that you already know about, with your own estimate attached.

The ten areas and the evidence to prepare

Area Question the reviewer asks Evidence to have ready Red flag
IP and ownership Does the company own all code? Signed IP assignment from every employee and contractor; repo ownership in a company org Repos in a freelancer's personal account
Open-source licenses Any copyleft or unknown licenses? License inventory for PHP and JS dependencies AGPL or unlicensed packages in a closed SaaS without review
Architecture Can it grow without a rewrite? One-page diagram, data flow, list of services Nobody can draw it
Dependencies How far behind is the stack? Framework and runtime versions, upgrade history Unsupported PHP or Node versions in production
Security Known vulnerabilities, leaked secrets? Audit output, secret scan, pentest summary Live API keys in git history
Infrastructure What does it cost and who controls it? Cloud accounts in company name, monthly bill, IaC or setup docs Servers on a developer's personal card
Data and privacy What personal data, where, how protected? Data inventory, retention policy, DPA list Production data copied to staging unmasked
Reliability Can you recover from a bad day? Backup policy, last restore test, incident log Backups never restored
Delivery How does code reach production? CI pipeline, deploy frequency, rollback steps Manual FTP deploys
Team Who knows what? Contributor map, docs, onboarding guide One person owns most of the critical code

A blurred server room aisle
AI-generated image.

Running the checks yourself before they do

You can produce most of that evidence in an afternoon. These are the commands I run on a typical Laravel + Vue codebase.

1. License inventory

# PHP dependencies: name, version, license
composer licenses --format=json > dd/licenses-php.json

# JavaScript production dependencies
npx license-checker --production --summary
npx license-checker --production --json > dd/licenses-js.json
Enter fullscreen mode Exit fullscreen mode

Read the summary for anything outside the permissive family (MIT, BSD, Apache-2.0, ISC). GPL-family and AGPL packages are not automatically a problem, but each needs a sentence explaining how you use it. "UNKNOWN" entries need an answer before the reviewer finds them.

2. Known vulnerabilities

composer audit --format=plain
npm audit --omit=dev
Enter fullscreen mode Exit fullscreen mode

Do not try to reach zero overnight. Export the output, mark what is fixed, what is not reachable in your code, and what is scheduled. A dated triage note reads far better than a clean report that was obviously produced the day before.

3. Secrets in git history

gitleaks detect --source . --log-opts="--all" --report-path dd/gitleaks.json
Enter fullscreen mode Exit fullscreen mode

This scans every commit, not just the current tree. A key deleted two years ago is still in history and still valid if nobody rotated it. The fix is rotation first, history rewriting second (and only if you understand the impact on every clone). I walked through the wider handover side of this, including secret inventories and runbooks, in Clean Codebase Handover: Runbooks, Secrets, Zero Lock-In.

4. Stack versions in one screen

php artisan about
node --version && npm ls vue nuxt --depth=0
Enter fullscreen mode Exit fullscreen mode

php artisan about (Laravel 9.21 and later) prints the Laravel and PHP versions, environment, cache and driver settings. Compare them with the official support windows and write down the upgrade path for anything close to end of life.

5. Key-person risk from git

# Commits per author over the last 12 months
git shortlog -sn --no-merges --since="12 months ago"

# Who touched billing code most
git shortlog -sn --no-merges -- app/Billing app/Http/Controllers/Billing
Enter fullscreen mode Exit fullscreen mode

Commit counts are a rough signal, not a performance metric. What reviewers look for is concentration: if one name dominates payments, auth or deployment, they will ask what happens if that person leaves. The answer should be documentation and a second engineer who has shipped changes there.

6. Backups you have actually restored

A backup policy document is not evidence. A dated restore test is. Restore last night's database into a scratch environment, time it, and record the result. If you need a framework for the targets, SaaS Backup & Disaster Recovery: RPO, RTO and PITR covers how to set them.

7. Server hardening snapshot

Reviewers increasingly ask for a short infrastructure security summary: SSH access, firewall rules, patching, who has root. If you run on VPS servers, the items in Linux VPS Hardening Checklist for Laravel Servers map almost one-to-one to what they ask.

Structuring the data room folder

Keep the technical section boring and predictable:

dd/
  01-overview/architecture.pdf, stack-versions.txt
  02-ip/ip-assignments-index.xlsx, repo-ownership.txt
  03-licenses/licenses-php.json, licenses-js.json, notes.md
  04-security/audit-triage.md, gitleaks-summary.md, pentest-summary.pdf
  05-infrastructure/accounts.md, monthly-cost.xlsx, diagram.pdf
  06-data/data-inventory.md, retention.md, subprocessors.md
  07-reliability/backup-policy.md, restore-test-2026-11.md, incidents.md
  08-delivery/ci-pipeline.md, release-process.md
  09-team/contributors.md, onboarding.md
  10-known-issues/tech-debt-register.md
Enter fullscreen mode Exit fullscreen mode

The last folder matters most. A tech debt register you wrote yourself, with rough effort estimates, lets you frame the fix list before the buyer frames it as a discount.

Common failure modes I see

  • Contractor code with no paperwork. The most expensive finding, because it is legal, not technical. Collect assignments from past freelancers while they are still reachable.
  • Cloud accounts under personal emails. Move domains, DNS, cloud and app-store accounts into company-owned identities with at least two admins.
  • A rewrite "in progress". Two half-systems look worse than one old one. Either finish, freeze or describe it as a strangler migration with clear milestones.
  • Costs nobody can explain. Be able to say what each line on the cloud bill does and what it would cost at twice the traffic.
  • Production data in staging. It turns a privacy question into a breach question. Mask it.

30-day preparation framework

  1. Week 1: Ownership. IP assignments, repo and account ownership, domain registrar.
  2. Week 2: Scans. Licenses, vulnerabilities, secrets, versions; triage each.
  3. Week 3: Proof. Restore test, architecture diagram, cost sheet, deploy walkthrough.
  4. Week 4: Narrative. Tech debt register with estimates, key-person plan, data room assembled.

FAQ

How long does technical due diligence take?

It depends on the deal size and how prepared you are. Prepared companies with an organised data room move quickly; the slow cases are almost always caused by missing ownership paperwork or answers that need a developer to dig.

Do investors read the source code?

Usually they sample it. Reviewers look at structure, tests, a few critical paths like auth and billing, and run automated scans. They care more about risk and maintainability than style.

Is using GPL or AGPL open source a deal-breaker?

Not automatically. It depends on how the code is used, modified and distributed, and AGPL has network-use terms that matter for SaaS. Document each case and get legal review for anything copyleft in your core product.

What if our framework version is out of support?

Expect it to be a price adjuster rather than a deal-breaker, provided you show an upgrade plan with an estimate. Unsupported runtimes without a plan read as unpriced risk.

Should we fix everything before the review?

No. Fix deal-breakers, rotate leaked secrets and run a restore test. Document everything else honestly in a tech debt register. Hidden issues found by the reviewer cost more trust than known issues you disclosed.

Who should own diligence preparation internally?

One technical lead with authority to get answers, supported by whoever handles legal and finance. Splitting it across many people produces inconsistent answers, which reviewers notice.


Follow & Connect

Watch on YouTube: Resume Boost AI · PixelValut · Daily AI News By Ali

Top comments (0)