Sample report

What the other side's engineers would find, on a real product.

This is the complete Full Audit of nicuk/docsflow, one of our own products: a multi-tenant platform where businesses ask questions of their own documents. Nothing is edited or hand-picked. It is what the audit returned, and every finding points at a file and line you can open on GitHub.

Scanned
September 22, 2026
Codebase
50,238 lines
Files read
160 of 477
Ranked risks
2 critical · 5 high · 4 medium

What you're looking at

The automated Full Audit, exactly as a buyer receives it. Start with the Snapshot, then the Action Plan: it groups every fix by when it matters, including the ones to close before you raise or sell.

What the due-diligence review adds

A senior engineer checks these findings against the code, removes anything that doesn't hold up, adds what automation can't judge, and signs a written summary. That review isn't shown here.

Your system at a glance

Drag nodes•Scroll to zoom•Click for details
  • Critical risk: 0
  • High risk: 0
  • Medium risk: 7
  • Low risk: 24

Who holds the knowledge

3 critical areas depend on one person.

AreaCommitsPeopleTop person's shareReading
migrationscritical71100%One person
app/authcritical21100%Too few commits to judge
app/logincritical11100%Too few commits to judge
docs/databasecritical41100%Too few commits to judge
lib/authcritical41100%Too few commits to judge
supabase/migrationscritical101100%One person
app/api/authcritical251100%One person
app/api/stripecritical41100%Too few commits to judge

Measured from commits in the last 12 months, not from understanding. Squash merges, shared accounts, and AI agents committing under a person's name can all make one person look like many, or many look like one.

The Bottom Line

This is a well-architected document-question-answering product for business customers, with genuinely strong engineering in its core search and reliability systems. However, the administrative side of the product currently trusts information that a visitor can simply type in themselves, meaning someone who knows a customer's account number and an administrator's email could read that customer's user list or send out invitations. That is a small, fixable set of problems — roughly a week of work — but it should be handled before any enterprise security review or investor technical audit.

Several issues need attention soon. Plan for 2-4 weeks of focused development work.

If Nothing Is Done

The administrative doors on your platform are effectively unlocked for anyone who knows two pieces of ID information, and one of your invitation systems hands out access codes that are guessable. If a customer discovers this — or an enterprise buyer's security team finds it during a review — you lose the deal and potentially the customer, and you spend weeks on emergency response and breach notification instead of building product. Separately, with no automated safety checks running before code goes live, every change your team ships is a coin flip, which quietly slows down every future release.

Critical Alert2 critical

2 critical and 5 high-severity vulnerabilities detected across 11 total issues + 3 high-priority quality issues

Feature Coverage

5 of 11 features have test coverage

6 features need automated verification

45%

Coverage

5 verified6 unverified
71/100

System Health Score

What the score means

70 – 100Strong
45 – 69Gaps present
25 – 44Needs attention
10 – 24Critical state

Critical risks cap the max score

AI Code Quality: 66%

Your system is well-built and maintainable

Risk Overview

11 risks + 13 quality issues
Critical
2
High
5
Medium
4
Low
0
3 high-priority quality issues

Want this for your own codebase?

Scan it free in about a minute, or ask for the human-reviewed version before a raise, sale or acquisition, from $1,500.