Code Security & Quality (SAST/SCA)

As of 19 Nov 2026, the panel rates SonarQube (3.5/5), Aikido Security (3.5/5) and Endor Labs (3.5/5) highest in Code Security & Quality (SAST/SCA), out of 19 products evaluated. Ratings that look equal are rounded to half stars; the order uses the unrounded rating, then the name. The highest-rated European option is SonarQube (EU origin, foreign-owned, Switzerland). Of the products whose own pages state a free plan, the panel rates SonarQube highest. Ratings are the AI panel’s opinion, formed from public evidence: how scores are made.

The bench · 6 judges, built for this category Meet the bench ↓ Fold ↑

The Lead Developer v1.0

Lead developer at a web agency in Cologne with fourteen developers building Laravel, TypeScript and React Native projects for clients on GitHub and GitLab. Wants every merge request checked for bugs, injection flaws and vulnerable packages before review, comments in the merge request rather than a separate dashboard, the same rules in VS Code and PhpStorm, few false positives, and a bill that does not jump each time a freelancer pushes a commit.

The Application Security Engineer v1.0

Application security engineer at a payments company in Frankfurt with 120 developers, two of them security champions per team, under PCI DSS and DORA. Wants taint analysis that follows user input across files, reachability so the backlog holds only vulnerable code that is actually called, secrets checked for validity across the whole git history, push protection, custom rules for in-house frameworks, and reports an auditor accepts.

The Engineering Manager v1.0

Head of engineering at a B2B SaaS company in Munich with six teams and a ten-year-old Java and Kotlin codebase. Wants technical debt and code health shown per team and hotspot, a quality gate on new code that every team keeps, trends to show the board that the rewrite budget worked, findings routed to Jira, single sign-on and roles per team, and to know what the AI-written code the teams now merge does to the numbers.

The Platform Engineer v1.0

Platform engineer at an insurer in Hanover who runs a self-managed GitLab and Jenkins for 300 developers inside the company network; source code may not leave it. Wants a self-hosted edition or a scanner that runs in the pipeline and sends only results, SAML and SCIM from Entra ID, an API to roll scan settings out to 900 repositories, SARIF output, audit logs, and upgrades that do not break the pipelines.

The Product Security Officer v1.0

Product security officer at a machine builder in Baden-Württemberg whose controllers and service apps fall under the EU Cyber Resilience Act. Needs an SBOM per release in a standard format, every component's vulnerabilities and licences tracked after shipping, evidence of secure development an auditor can file, and to know where the firmware source code is stored, for how long, and whether any AI model is trained on it.

The Skeptic v1.0

Has seen "zero false positives" bury a team under four thousand findings on day one, a "free for open source" plan that counted every bot as a contributor, and an "AI autofix" that shipped the repository to a model in the US. Asks how contributors are counted, which languages get real data-flow analysis rather than regular expressions, what happens to the code after a scan, whether the findings export when the contract ends — and what the second year costs.

Every judge is AI, versioned, and scores independently against the same published anchors — disagreement is printed, not averaged away.

Head-to-head

Evaluated 19

Listed alphabetically by default. Sorting by rating or sovereignty is your choice and follows the published method. Scores are the panel’s opinion, formed from public evidence. How scores are made →

Aikido Security

EU-Made Panel rating

Arnica

Rest of world Panel rating

Checkmarx One

Provenance unknown Panel rating

Codacy

EU-Made Panel rating

CodeScene

EU-Made Panel rating

DeepSource

Rest of world Panel rating

Endor Labs

Rest of world Panel rating

GitGuardian

EU-Made Panel rating

GitHub Advanced Security

Rest of world Panel rating

Kiuwan Code Security

Provenance unknown Panel rating

Mend.io

Provenance unknown Panel rating

Qlty

Rest of world Panel rating

Qodana

EU-Made Panel rating

Semgrep

Rest of world Panel rating

Snyk

Rest of world Panel rating

SonarQube

EU origin, foreign-owned Panel rating

Sonatype Lifecycle

Provenance unknown Panel rating

Veracode

Provenance unknown Panel rating

Xygeni

EU-Made Panel rating

Questions about Code Security & Quality (SAST/SCA) software

What is the best Code Security & Quality (SAST/SCA) software?

As of 19 Nov 2026, the panel rates SonarQube (3.5/5), Aikido Security (3.5/5) and Endor Labs (3.5/5) highest of the 19 Code Security & Quality (SAST/SCA) products evaluated. Ratings that look equal are rounded to half stars; the order uses the unrounded rating, then the name. The ratings are the AI panel’s opinion, formed from public evidence. Each judge weighs the criteria differently, so read the scores for the criteria that matter to you.

Which Code Security & Quality (SAST/SCA) software is European?

As of 19 Nov 2026, 7 of the 19 Code Security & Quality (SAST/SCA) products evaluated are European by provenance: SonarQube, Aikido Security, Qodana, Codacy, GitGuardian, Xygeni and CodeScene. The highest-rated of them is SonarQube (EU origin, foreign-owned, Switzerland).

Which Code Security & Quality (SAST/SCA) tools host data in the EU?

Per the vendors’ own pages as captured, SonarQube stores customer data in the EU by default, though not exclusively. A product not named here may still host in the EU; its pages as captured did not prove it.

Which Code Security & Quality (SAST/SCA) tools have a free plan?

As of 19 Nov 2026, the vendors’ own pages state a free plan for SonarQube, Aikido Security, Qodana, Semgrep, GitHub Advanced Security, Sonatype Lifecycle, GitGuardian, Snyk, and 3 more (see the full list). Each product page links the pricing page it was read from.

How were these Code Security & Quality (SAST/SCA) tools evaluated?

Each product is scored by a bench of 6 AI judges built for this category, against the same published rubric of 7 criteria, using public evidence captured from the vendor’s own pages. Every score is AI-generated; nobody tested the products hands-on, and no score can be bought. The full method is on the methodology page.