S/08audit and optimisation

Code and architecture audit before you buy or inherit a product

A code and architecture audit is an independent review of software you did not write — before you acquire it, take it over from another vendor, or spend another six figures on it. One engineer reads the codebase, the database schema and the infrastructure, profiles what is genuinely slow, checks the security posture, and delivers a written report that ranks every finding by risk and by the hours it takes to fix.

Discuss your taskFree scoping call, scope and price within 1 day, reply within 2 hours

What this solves

You are buying a product on trust

The seller demos the happy path and tells you the code is clean. An audit gives you a second opinion before the wire goes out: what is actually in there, what it costs to bring to standard, and which findings are grounds to renegotiate.

The agency is gone, the code is not

No documentation, no author, no idea where anything lives. We reconstruct the structure, the data model and the path to production, then mark the parts of the system that are dangerous to touch without preparation.

It is slow and nobody knows why

"Slow" almost always resolves into three or four specific queries and one screen that was designed wrong. We profile the database and the application under realistic load and show you where the seconds actually go.

Nobody has checked the security posture

How passwords and tokens are stored, who can read whose data, which admin panels and ports face the internet, what secrets sit in the git history. Every finding carries a risk level, not a vague "should be hardened".

The quote does not match the work

A $3,000 quote for something that smells like forty hours is a normal reason to call us. We read the requirement technically and tell you what it is worth in hours before you sign anything.

Every change breaks something else

When editing a price in one module breaks reporting in another, the problem is module boundaries and missing tests, not the developers. We show where the coupling is unnecessary and the smallest set of tests that stops the regressions.

How the work runs

  1. Free scoping call and access

    1 day

    We start with what the audit is for: an acquisition, a dispute with a vendor, or production being slow. Then we need read-only repository access, a database dump or replica, and view access to monitoring and hosting.

  2. Reading the code and the architecture

    2–4 days

    A human reads the code, not only the linters: module boundaries, data model, migrations, error handling, test coverage, commit history. We flag every area that exactly one person in your company understands.

  3. Performance profiling

    1–2 days

    Slow queries pulled from pg_stat_statements and EXPLAIN, application profiling, key flows replayed under load. Every "it's slow" turns into a named query, a missing index or an N+1.

  4. Security and infrastructure review

    1–2 days

    Authentication and authorisation, secrets in the repository, dependencies with known CVEs, exposed ports, backups and an actual restore test, CI/CD. We work down the OWASP ASVS checklist.

  5. Report and walkthrough

    1–2 days

    You get the written report — findings by risk level, the consequence of leaving each one, an hour estimate per fix — plus an hour on a call going through it, so the list is usable by whoever ends up doing the work.

  6. Remediation, if you want it

    from 5 days

    If we do the fixing, we take the top of the list and work hourly or on a monthly retainer. Everything we ship carries a one-month warranty.

Stack

Tooling follows the project's stack — auditing Node/Postgres and auditing PHP/MySQL are different routes through the same checklist. Below is what comes up most often; the report itself is written by an engineer, not exported from a linter.

  • PostgreSQL: EXPLAIN ANALYZE, pg_stat_statements
  • Node.js profiling: clinic.js, 0x
  • Load scenarios: k6
  • Lighthouse and Core Web Vitals
  • Semgrep, npm audit, Trivy
  • OWASP ASVS checklist
  • Sentry, Prometheus + Grafana
  • Docker and docker compose
  • GitHub Actions, GitLab CI
  • TypeScript strict mode, ESLint
  • Git history: hotspots and bus factor

What it costs

Express audit

$600 – $1,200 / 60 000 – 120 000 ₽

20–40 hours. Repository and infrastructure end to end, no deep profiling. You get a map of the system and the ten findings that actually matter, each with an hour estimate. Enough to decide buy or walk away.

Full audit

$1,500 – $3,000 / 150 000 – 300 000 ₽

50–100 hours. Code, database schema, performance under load, security checklist, CI/CD and a backup restore test. Full report with priorities and hours per item, plus a walkthrough call. Codebase size and service count move the number.

Audit plus remediation

from $2,500 / from 250 000 ₽

The audit plus work on the top of the list: indexes and queries, bottlenecks, critical security holes, a deployment path that stops being scary. After that, an optional retainer from $400 / 40 000 ₽ a month with a dedicated engineer.

Prices are indicative and do not constitute a public offer. The final cost, scope and timeline are fixed in a contract or invoice after we have discussed the task.

Related case study

The audit checklist comes out of systems we run ourselves. KMD is a CRM we built from scratch for a large HoReCa service company: 12+ modules, 89 database migrations, access split across 9 roles and 87 permissions. Those are exactly the places we open first in someone else's product — the permission matrix, the migration history, the seams between modules.

Read the KMD case

Frequently asked questions

How much does a code audit cost?
An express audit is $600–$1,200 (60 000–120 000 ₽) and a full audit is $1,500–$3,000 (150 000–300 000 ₽). Everything derives from a $30/hour rate, so the price is simply scope: codebase size, number of services, how deep the security review goes. Payment is 50% upfront, 50% on acceptance of the report.
How long does an audit take?
Three to five working days for the express version, seven to twelve for the full one. The clock starts when access is in place: read-only repository, a database dump or read replica, and monitoring. In practice access is what delays an audit, not the work.
What do I actually get?
A written report and an hour on a call walking through it. The report contains a map of the system, findings ranked by risk with the consequence of ignoring each one, an hour estimate per fix, and a two-week plan for the top of the list. It is written to be handed to any vendor, not only to us.
Will you sign an NDA, and what access do you need?
Yes, the NDA is signed before anything is shared. Access is read-only: a repository invite, a database dump or read replica, and view-only monitoring and hosting dashboards. We never need write access to production for this work, and we ask you to revoke everything the day the report lands.
Can I share the report with an investor or an acquirer?
Yes, and that is one of the main reasons people order it. The report states what we checked, what we could not check and why, so a third party can judge how much weight to give it. There are no invented scores — each finding points at a file, a query or a configuration line.
Is this a penetration test?
No. We review code and configuration from the inside: authentication, authorisation, secrets, dependencies with known CVEs, exposed services, backups and whether they really restore. A penetration test attacks the system from the outside; if that is what you need, we will tell you rather than sell you the wrong thing.
Do you fix what you find, or only report it?
Either, and you decide after reading the report. Plenty of clients hand it to their own team — that is a fine outcome and the document is written for it. If we do the work, it is $30/hour or a retainer from $400 a month with a dedicated engineer, and anything we ship carries a one-month warranty.
What if the audit finds nothing serious?
Then you get that conclusion in writing, with the evidence behind it. It is a real result: it settles an argument with your current vendor, or confirms that the product you are buying is safe to take on. The price does not depend on how many problems we find — you pay for the check, not for bad news.

Send us the repository link

The scoping call and the estimate are free, with no obligation. We reply within two hours during business hours and tell you honestly whether you need the full audit or the express one.

Discuss your task
Code and architecture audit — Corexy