HomeProduct

A pentest that proves every finding.

Nullgap reads your repository the way an attacker would, then attacks each lead in a sealed sandbox. You get only what really works, with the requests that prove it.

FoundProvenFixed✓ Proven fixed
A sealed sandbox box with a padlock, examined by a magnifier, stamped Proven
How it works

From repository to proven findings in four steps.

  1. Connect a repository

    Sign in with GitHub and pick a repository. Nullgap counts the code and shows you the exact price before anything runs.

  2. Find the leads

    Nullgap AI reads the code like an attacker: routes, queries, auth checks, secrets and dependencies. Each lead is a claim, not yet a finding.

  3. Prove them in a sandbox

    Every lead is attacked in a sealed sandbox built from your code, with no outside network. Only leads that pass all four checks become findings.

  4. Fix, and prove the fix

    Each proven finding gets a fix pull request. The original exploit runs again against the fix, and it has to fail.

What you get

A short list of real problems, each one replayable.

Findings arrive in the console with the proof attached. Leads that were tested and ruled out are listed too, so you know what was covered.

Findings for a sample repository
FindingWhereState
Any signed-in user can read any invoicesrc/routes/invoices.ts:3Proven
SQL injection in the report filtersrc/reports/query.ts:41✓ Proven fixed
Password reset token never expiressrc/auth/reset.ts:18Fixed
6 other leads testedShown not to be exploitableRuled out
Proof · Any signed-in user can read any invoice
attack   GET /invoices/1002 as customer A (owner: B)  → 200, B’s invoice   Worked
repeat   same request, fresh session                 → 200, B’s invoice   Worked
control  A opens A’s own invoice                     → normal             No effect
check    independent review of the evidence                              Accepted
Illustrative sample. Every report also includes an executive summary, an attestation letter and an evidence bundle per finding.
Key details

What to expect, including the limits.

Priced by code size, shown first
$750 up to 50k lines, $2,500 up to 250k, $6,000 up to 1M, and a quote above that. You see the exact price before it runs.
Live testing included
Add a staging address and up to five test accounts to prove access control across users in the same pentest.
Free retest within 60 days
Run the same repository again after you fix, at no charge, within 60 days.
Refunds when we fail
If a run fails on Nullgap’s side, it is refunded. Cancelled runs are charged only for work done, never more than the price.
Resume after a stop
A stopped pentest can be resumed from where it left off instead of starting over.
Very large repositories
Repositories over 4 GB, when added by address, are refused at once with a clear message, before anything is charged.
Your code is not kept
The repository is checked out for the run and deleted when it ends. AI prompts and responses are never stored.
The proof guarantee
If a pentest proves nothing, your next pentest of that repository is half price.
Related

Pentest questions

What counts as proven?

The attack works, it works again, a harmless control run doesn’t, and an independent check accepts the evidence. A status code alone never counts. How proof works.

Does the pentest touch production?

No. Proofs run in a sealed sandbox built from your code, with no outside network. Live testing only reaches staging addresses on your scope list.

What if nothing is proven?

Nothing is reported as a finding, and you still see the leads that were tested and ruled out. Your next pentest of that repository is half price.

Can our auditors use the report?

Yes. Each pentest produces an attestation letter, an executive summary, a full report and an evidence bundle for every finding.

✓ Proven fixed

Get your first proof today.

Connect a repository and watch Nullgap find, prove and fix. You only ever see what it could prove.

Free 14-day trial · No card · or book a demo with a security engineer