The logic bugs
that pass your linter.
Faultmark reads your whole codebase for the logic and security bugs your scanners miss, verifies every finding against your actual source, and throws away anything it cannot prove. You only pay for the bugs you keep.
A reading in three passes.
Point us at the code
Give us read access to one repository, or name a public one. No writes, no CI changes, no standing access.
We find, then verify
Every candidate is re-read against your actual source. If we cannot point to the exact line, we delete it before you see it.
You decide
You get a report of real bugs only, each with a concrete fix, to file and line. You merge what you want. Nothing is auto-applied.
This is what a review looks like.
A short walkthrough of the actual decisions you make, from repository selection to a review-ready pull request.
REPOSITORY
Choose the code you want reviewed.
Faultmark asks for a repository, a branch, and your approval before it starts.
We delete our own findings.
Most tools show you everything the model thought it saw. We don’t.
A first pass reads the whole codebase and proposes candidate bugs, each with the exact code as evidence.
A second, adversarial pass re-reads each candidate against your real source and tries to disprove it.
If the quoted code is not in your file, or a guard covers the path, the finding is deleted. You never see it.
Evidence required
A finding has to quote the exact line from your source. No quote, no finding.
Adversarial re-check
A separate pass tries to disprove every candidate before it ever reaches your report.
Short by design
Only findings that survive verification make the report. That is why the report is short.
Everything a senior engineer would catch. Automated.
The whole codebase, not the diff
We read how values flow across files, not just the lines that changed in a pull request.
Verified against your source
Every finding is checked against your real code. Anything we cannot prove is dropped, not shown.
We show what we dropped
The report says how many candidates we discarded and why. Precision you can check, not a number you have to trust.
A concrete fix, not a hint
Every finding comes with a patch, to file and line. You review it and merge. Nothing auto-deploys.
You only pay for what you keep
A bug counts when you merge the fix or confirm it is real. If your scanners already caught everything, you owe nothing.
From finding to pull request
When you accept a fix, it becomes a branch and a pull request for you to review. Faultmark never merges.
The work that happens after the first scan.
Manual scans build deep repo context. Push and pull-request audits use that context automatically, while developers are already working in GitHub.
Run it yourself, or have us do it.
Free
for solo developers- ✓Scans3 / month
- ✓Findings verified against sourceincluded
- ✓Fix proposalsincluded
Pro
for teams that shipSELF-SERVE- ✓Scansmonthly budget
- ✓Automatic push & PR auditsincluded
- ✓Findings verified against sourceincluded
- ✓Fix proposalsincluded
The Founding Audit
one repository, one timeYou only pay for the bugs you keep.
- ✓We read one repository and send you every real bug we find, ranked, each with a concrete fix.
- ✓Read access only. We never open a pull request, never touch your CI, never ask for write access.
- ✓Security findings come to you privately, never a public issue.
- ✓A bug counts only when you merge the fix or confirm it is real. If we find nothing you keep, you pay nothing.
