What triggers a finding
- SonarQube
- A scheduled scan, a CI "sonar:scan" step, or a pull-request decoration
- CodePatrol
- A push or pull request on a connected repository
Product comparison
Different windows, different outputs. SonarQube full-scans the codebase on a schedule; CodePatrol reads the commit your team just pushed and opens a finding on the diff — and that does not make them substitutes.
Try CodePatrol
Start free with one repository and 100 commits a month, or compare the plans before you begin.
CodePatrol and SonarQube both produce code findings, and both end up on a list of "static analysis tools we could install." They do not, however, inspect the same window. SonarQube runs a deep, scheduled, whole-codebase scan against a quality profile your team authors. CodePatrol reads a PR diff the moment the team pushes it and opens an issue only when a rule actually matches. They are not the same product, they do not compete for the same budget line, and the right answer is often "use both."
The most useful way to read the difference is the trigger — what makes each tool open a finding, and what does it read when it does.
| Dimension | SonarQube | CodePatrol |
|---|---|---|
| What triggers a finding | A scheduled scan, a CI "sonar:scan" step, or a pull-request decoration | A push or pull request on a connected repository |
| When it fires | On a cadence the team configures — nightly, per-PR, or both | On every push — the same instant your CI would see the diff |
| What it inspects | The whole codebase against a quality profile of vendor and custom rules | The actual lines the team added, removed, or refactored |
| Output | A dashboard of findings across the repo, plus per-PR decorations when configured | A code-review finding on the diff (GitHub Issue on paid plans, in-context PR review) |
| Languages | Twenty-plus, including JVM, .NET, JS/TS, Python, Go, C/C++, and more | Focused on the languages commit-time AI detection covers best — JS/TS, Python, Go, JVM-adjacent stacks |
| Rule authoring | Authored by your team against a quality profile; vendor rules shipped as plugins | Curated rule set maintained by CodePatrol; custom rules available on the Team tier |
| Output channel | Dashboard + IDE plugin + per-PR decoration in CI | PR review comment on the diff, escalating to a GitHub Issue on confirmed findings |
SonarQube’s substrate is the full repo. The quality profile walks every file on every scan, the dashboard surfaces findings the team has not yet triaged, and the per-PR decoration is a thin slice of that whole-repo surface. CodePatrol’s substrate is the diff. The tool reads the lines the team actually shipped, evaluates them against a curated rule set, and decides whether those lines introduce a vulnerability pattern or a regression. The two tools answer different questions about the same repository.
Both products do detect vulnerabilities and code smells — that is the visible overlap. The overlap is real, and the divergence lives in the trigger and the alert shape, not in the words "we scan code."
For vulnerabilities, the SonarQube taint analyzer walks dataflow across the entire repository to find sources and sinks; on a substantial service, that scan can take minutes per run and the team usually gates it to nightly, because blocking PRs on a ten-minute scan is not realistic. CodePatrol evaluates the lines the team changed that day, against a curated set of insecure-by-construction patterns, in seconds. The review window is where the two products trade off: SonarQube sees more code, but sees it later; CodePatrol sees less code, but sees it before merge.
On code smells — duplication, complexity thresholds, naming conventions — SonarQube has years of accumulated rules and a quality profile your team tunes by hand. CodePatrol does not posture as a code-smell platform. The product’s posture is "did we ship a vulnerability," not "is this service too complex." A team that needs active code-smell management will configure SonarQube for that surface; it is a feature CodePatrol does not compete on.
Both products cover security, but they cover it at different depths with different output shapes. SonarQube ships a large vendor-managed ruleset — OWASP, CWE, language-specific taint analyzers — that the team tunes through a quality profile over time. CodePatrol ships a smaller, curated rule set scoped to the patterns the team needs to catch before merge; the rules are maintained by the vendor rather than authored per-project.
CodePatrol opens a GitHub Issue only when a rule actually matches. The GitHub Issues (Write) scope is gated to paid plans, and we deliberately surface a confirmed finding — not a heuristic — so the tickets your team triages are the tickets that need a human. Every Issue is the output of a real rule evaluation on the diff the team shipped, not a noisy pattern sweep.
SonarQube’s output shape is a dashboard with thousands of pre-existing findings plus a smaller delta on each scan. CodePatrol’s output shape is a small handful of confirmed findings per week on the lines the team actually changed. Both are correct behaviors — the products are answering "is there anything in this codebase?" vs. "did we ship a bad change?" — and a team that needs both views will run both. The two products cover security at different depths: SonarQube covers the whole-codebase surface and the team’s existing debt; CodePatrol covers the PR-time window where the team can still act on the finding before merge.
Deployment model is the second cleanest signal that the two products sit on different floors. SonarQube has two shapes — the Community Edition, which is a self-hosted Java service your team operates on a VM or in Kubernetes, and the SonarCloud edition, which is the same product hosted by SonarSource. CodePatrol is a hosted SaaS only; the team does not operate any infrastructure to run the product.
| Dimension | SonarQube / SonarCloud | CodePatrol |
|---|---|---|
| Deployment model | Self-hosted CE/Developer/Enterprise, or hosted SonarCloud | Hosted SaaS only |
| Infra footprint | JVM service, a backing database (Postgres or similar), object storage for branch caches — your team operates all of it on CE | None — provided by the platform |
| Ops burden | Upgrades, scanner plugin maintenance, database migrations, capacity tuning | None — the platform handles upgrades and capacity |
| Data residency | Self-hosted keeps the data inside your perimeter; SonarCloud is vendor-hosted | Vendor-hosted; data residency posture is whatever the platform offers |
| Air-gapped / regulated CI | Self-hosted CE is the right answer | Not available — requires outbound connectivity to the platform |
A regulated environment with a hard data-residency requirement will reach for self-hosted SonarQube — that is the deployment shape the product was built to support, and the cost is the operations burden the team takes on. A team that does not need air-gapped CI will look at hosted SonarCloud or hosted CodePatrol and pick based on which window (scheduled full repo vs. commit-time diff) matters more to the security posture.
The two products price against different denominators, which is the third cleanest signal that they cover different floors.
| Cost dimension | SonarQube / SonarCloud | CodePatrol |
|---|---|---|
| Pricing denominator | Per line of code (SonarCloud developer plan), or per instance capacity for self-hosted | Per repository, with a free one-repo tier |
| Pricing axis | LOC-based: more code → more cost; free tier is 10k LOC across languages | Repo-based: cost tracks the number of connected repositories |
| Public / open-source repos | Free on SonarCloud | Free tier available — one repo, 100 commits / month |
| Small private team (5 engineers) | SonarCloud developer plan charges per LOC band; LOC grows with the codebase | Pro at $49/month covers up to 10 repositories with unlimited commits |
| Larger organization | Per-LOC growth on hosted; self-hosted capacity-planning on-prem | Team tier adds custom rules and ticket-routing SLAs at per-repo pricing |
| Free-to-paid upgrade path | Hit the free LOC cap, then upgrade plan | Yes — free tier hits its commit cap, Pro is a single upgrade |
A five-engineer team evaluating what they actually pay sees different shapes on each line item. SonarCloud’s per-LOC pricing model grows with the codebase; CodePatrol Pro is $49/month for up to ten repositories. Neither pricing model is wrong — the products are pricing against the unit they affect (lines of code vs. repositories). A team should pick based on which axis their security budget line items fall on.
There is a clean rule of thumb for which tool fits which problem. SonarQube is the right answer when the dominant pain is "what is in our codebase, how bad is it, and what is the trend?" CodePatrol is the right answer when the dominant pain is "did we ship a vulnerability in this PR?"
The "use both" case is real and easy to justify. SonarQube on the codebase gives the team a quality gate and an existing-debt view; CodePatrol on PRs gives the team a same-day review on the lines they changed. The two products do not collide — they cover different floors — and the total cost is a small per-LOC band plus a flat per-repo tier.
CodePatrol and SonarQube are not substitutes. If your pain is "we shipped a bad code change and want a reviewer before merge," CodePatrol is the right tool — it inspects the diff your team actually wrote and posts a confirmed finding on the PR. If your pain is "our codebase has thousands of existing findings, we need a quality profile, and we want a dashboard that trends over time," SonarQube is the right tool — it covers the whole-codebase surface and the existing-debt floor SonarQube was built to cover. The two products inspect orthogonal surfaces, and most teams end up running both. If you need both, use both — they do not compete; they cover different floors.
What CodePatrol explicitly does not do: replace a full SAST platform. We do not run a scheduled codebase scan, we do not trend findings across months, and we do not posture as a quality profile your team authors by hand. What SonarQube explicitly does not do: review your PR diff at commit time. It does not open a confirmed finding on the lines your team just shipped, and it does not own the PR-time review window where the team can still act on a vulnerability before merge.
Try CodePatrol
Start free with one repository and 100 commits a month, or compare the plans before you begin.