Skip to Content
Comparisons

How xgrep compares

How xgrep compares to other code security scanners. A single local binary that runs SAST, secrets, dependency and SBOM scanning on your machine, with no language runtime and without uploading your source code.

Most code-security scanners make you pick a side: a fast local linter that stays shallow, or a deep platform that wants your source in its cloud. xgrep is built to avoid the choice. Whatever tool you are comparing it against, three things stay constant.

One binary, installed in seconds

xgrep is a single self-contained executable. There is no Python or JVM to install, no package manager to satisfy, no plugins to resolve, no runtime to keep in sync. Download it, mark it executable, and scan. The same file works on your laptop, in CI, and inside an air-gapped build. See Installation.

Your code is analyzed on your machine

The first question to ask of any code scanner is whether it uploads your source to analyze it. Several do: Snyk Code sends your source to its cloud by default, and Semgrep's cross-file and AI features send code context to theirs. When the code you scan is the asset you are protecting, proprietary source, regulated data, or anything under data-sovereignty rules, that one answer should decide the tool before any feature comparison.

xgrep analyzes your source locally and only locally. Your code is never uploaded, and there is no server-side analysis mode that could receive it. Adding Mondoo Platform does not change that: the analysis still finishes on your machine, and only findings and an SBOM ever leave it, never the source. You get organization-wide posture without your code becoming the thing that travels to get it.

Two steps do use the network, and both send results rather than source:

  • Dependency vulnerability matching sends your SBOM, meaning package names and versions, to Mondoo Platform. There it is matched against Mondoo Vulnerability Intelligence, which covers CVEs and malicious packages across 50+ ecosystems. This needs a Mondoo account. Without one, xgrep still builds the SBOM and still reports SAST and secrets; only the vulnerability match is skipped. SBOM generation, the dependency inventory, and reachability analysis all run locally.
  • Secret validation (--validate, opt-in and off by default) probes a candidate credential against its own provider's API to see whether it is live.

A local tool, and a platform when you want one

xgrep is complete on its own, and it also integrates natively with Mondoo Platform. With a Mondoo account configured, the findings a scan produces publish automatically, while the code they came from stays put. Code, IaC, secrets, vulnerability and malware findings from every repository then sit in one place with risk scoring, exceptions and fix status. That is the wider software supply chain security picture, with xgrep as the scanner feeding it. Local text, JSON and SARIF output is identical either way, and --incognito keeps a scan entirely local. See Mondoo Platform.

xgrep is the code half of that picture. The other half is cnspec, which covers infrastructure as code, cloud accounts, containers, Kubernetes and running hosts, checked against policy and compliance frameworks. Findings from both land on the same asset, so you get code, build and runtime in one view instead of a SAST tool and a posture tool reporting separately.

One tool for the whole surface

A single xgrep run covers what usually takes several tools stitched together:

  • SAST with taint and dataflow analysis, not only pattern matching (code scanning)
  • Secrets, validated against the provider (secrets)
  • Dependencies, with reachability and vulnerability matching (dependencies)
  • SBOM, CBOM and AIBOM, software, cryptography and AI/ML bills of materials, from source
  • Code DSPM, what sensitive data the code holds and where it leaks (code DSPM)

Regulated and sovereign environments

Those bills of materials answer regulatory questions directly. An AIBOM inventories the models, AI SDKs, frameworks and inference runtimes a system actually uses, which is the component evidence behind the technical documentation the EU AI Act requires for high-risk systems under Article 11 and Annex IV. An SBOM answers the same question for software components, which is what the Cyber Resilience Act asks for. A CBOM inventories cryptographic assets, which is where a post-quantum migration plan starts.

xgrep produces all three from source, offline, with no account and nothing uploaded. That matters for sovereignty as much as for compliance, because the analysis runs inside whatever boundary you put it in, including an air-gapped one.

Where you want findings centralized, Mondoo Platform has deployment options to match: SaaS in an EU or US region, a single-tenant instance of your own, or a self-managed deployment in your own cloud account on Google Cloud, AWS or Azure, where the data never leaves your environment.

Accuracy you can check

On the OWASP Benchmark, xgrep finds every planted vulnerability with zero false positives, in Java and in Python. The per-tool pages below carry the rest.

Accuracy has two halves and a scanner has to win both. Recall is how much of what is really there it finds. Precision is how much of what it reports is real. Miss the first and the bug ships anyway. Miss the second and people stop reading the output, so the real findings are lost with the noise. Every number is measured on public, labeled corpora and reproducible from the tool versions we pin.

There is a third thing a scanner owes you: honesty about what it could not read. A parser never refuses a file — given input its grammar rejects, it recovers with error markers and returns a tree anyway, so a finding from a file the scanner only partly understood would otherwise look identical to one from a file it read perfectly. xgrep flags those files instead, so a finding drawn from a recovered parse is marked rather than silently mixed in with the rest. See output formats.

The comparisons

Last updated on

On this page