Pixee is your purple team.

It builds a threat model of your application and studies it the way an attacker would. Then it proves which vulnerabilities are real and opens the fix as a pull request.

Everything is signal

Your whole security program feeds Pixee.

Designs and tickets, code and pull requests, pen-test and bug-bounty reports, your own harnesses and the scanners you already run. Pixee checks them all against a threat model of your application, then sends back a verdict with the evidence behind it and the fix as a pull request.

Designs & plans
SpecsDesign docsPRDsTickets
Confluence, Notion, Google Docs, Jira, Linear
Code
RepositoriesPull requests
GitHub, GitLab, Bitbucket, Azure DevOps
Pen tests & bug bounty
Pen-test reportsTester notesBug-bounty reports
Custom harnesses & internal tools
Your own scannersCustom rulesInternal harnesses
Anything that emits SARIF
Scanners
SASTSCASecretsDAST
Native integrations
Pixeedesign · find · triage · fix
Back to your team
Design flagsThreat model & attack pathsVerdictsMerge-ready fixes
01pixee.design · Design review

Catch it in the design, before there's code to scan.

Pixee reads specs and tickets before code exists and flags the security promises they miss, while the fix is still an edit to the spec.

Foresight →
Spec · Password resetIllustrated
1. The user asks for a reset from the sign-in page.2. We email a reset link to the address on file.3. The link opens a form to set a new password.
Missing security promise

Reset links never expire

The spec sends a reset link by email but sets no expiry and no single-use rule.

Flagged in design review
02pixee.find · Threat modeling & detection

Before it looks for bugs, Pixee threat-models your application from the code.

Your scanners report findings one at a time. Pixee builds one model of how your application fits together, then follows the paths through it and shows you the ones that reach something an attacker wants.

One path through the threat model, illustrated
Trust boundaryAttack path
Your application
Entry point

Public API gateway

Passes client headers through to internal services.

Internal service

Billing service

Trusts the user header without checking who set it.

Sensitive asset

Admin actions

Any caller can act as any user, including an admin.

Critical
  1. 1Map

    Builds a threat model from your code, with its entry points, trust boundaries and sensitive assets.

  2. 2Follow

    Traces paths from an entry point across the trust boundaries and keeps the ones that end at a sensitive asset.

  3. 3Prove

    A verifier tries to disprove every lead. Pixee shows what it examined, what it found and what it ruled out.

Pixee Beacon →

03pixee.triage · Triage

Keep every scanner you run. Get a verdict on what they find.

Pixee investigates findings from your scanners, pen tests and bug bounty alongside its own, and writes down each verdict with the evidence behind it.

SASTSCASecretsDASTAny SARIF toolPen testsBug bounty
Finding · SQL injectionTriage
True positive

User input reaches the query in OrderRepository

No sanitizer on the path from the request parameter to the query string.

up to 95%

fewer false positives

Pixee customer cohort, 2025
04pixee.fix · Remediation

Real vulnerabilities come back as pull requests.

Pixee writes the fix in your conventions and opens a merge-ready pull request. Your developers review it like any other change.

Triage & Fix →
Pull requestFix
Ready to merge

Use a parameterized query in OrderRepository

− "… WHERE id = " + id
+ "… WHERE id = ?", id

Matches how OrderRepository already builds its other queries.

94%

end-to-end resolution rate, triage and fix

Pixee customer cohort, 2025
Show your work

What each agent records.

Each agent writes down what it found and decided, so you can see why.

pixee.designsecurity promises and the threat model from the design
pixee.findthe threat model from the code, attack paths, dismissed leads
pixee.triageverdicts and the evidence behind them
pixee.fixfixes and which ones your developers merged
MoneyGram

MoneyGram runs Pixee in its GitLab pipelines.

MoneyGram serves 60 million active customers across 200 countries and territories and nearly 500,000 locations.

96%

of findings processed automatically

6,500+

hours saved across triage and fixes

What is a purple team in application security?

A purple team puts attacker and defender work together: one side finds how an application can be broken, the other side fixes it. Pixee does both on your repositories. It threat-models your code, follows the attack paths, decides which findings are real and opens merge-ready fixes for your developers to review.

How does Pixee build a threat model from code?

Pixee reads your code and maps its entry points, trust boundaries and sensitive assets. It traces paths from an entry point across those boundaries and keeps the ones that reach a sensitive asset. A verifier then tries to disprove every lead, and Pixee shows what it examined, what it found and what it ruled out.

Does Pixee replace my scanners?

No. Keep the scanners you run. Pixee takes their findings through native integrations or any tool that emits SARIF, along with pen-test and bug-bounty reports. It investigates those findings next to its own, writes down each verdict with the evidence behind it, and cuts false positives by up to 95%.

Do my developers approve every fix?

Yes. Pixee opens each fix as a pull request written in your code's conventions, and your developers review it like any other change. Nothing merges on its own. Across the Pixee customer cohort in 2025, the end-to-end resolution rate was 94%, from scanner finding to a handled decision.

See it on a real repository

Watch Pixee work a repository end to end.

We'll walk through a real run: the threat model, what Pixee found, what it dismissed and why, and the fixes it wrote.

Book a demo
Self-hosted or in your VPCNothing merges on its own
Pixee94% and up to 95%: Pixee customer cohort, 2025 · 96% and 6,500+: MoneyGram case study