Security glossary

This glossary provides definitions for terms related to security features in GitLab. While some terms may have different meanings elsewhere, these definitions are specific to GitLab.

Analyzer

Software that analyzes a scan target type for security vulnerabilities. Internally, it is responsible for gathering required configuration parameters, performing necessary data transformations to convert the target into a standardized format for the scanner to execute the scan operation. Finally, it produces a report in the format required by the caller.

CI/CD-based analyzers integrate into GitLab using a CI/CD job. The report produced by the CI/CD-based analyzer is published as an artifact after the job completes. GitLab ingests this report, allowing users to visualize and manage found vulnerabilities. The generated reports adhere to the secure report format.

Many GitLab analyzers follow a standard approach using Docker to run a wrapped scanner. For example, the image semgrep is an analyzer that wraps the scanner Semgrep. However, some analyzers run directly within GitLab Rails or other target environments rather than in separate containers.

Attack surface

The different places in an application that are vulnerable to attack. Secure products discover and search the attack surface during scans. Each product defines the attack surface differently. For example, SAST uses files and line numbers, and DAST uses URLs.

Component

A software component that makes up a portion of a software project. Examples include libraries, drivers, data, and many more.

Corpus

The set of meaningful test cases that are generated while the fuzzer is running. Each meaningful test case produces new coverage in the tested program. You should re-use the corpus and pass it to subsequent runs.

CNA

CVE Numbering Authorities (CNAs) are organizations from around the world that are authorized by the Mitre Corporation to assign CVEs to vulnerabilities in products or services within their respective scope. GitLab is a CNA.

CVE

Common Vulnerabilities and Exposures (CVE®) is a list of common identifiers for publicly known cybersecurity vulnerabilities. The list is managed by the Mitre Corporation.

CVSS

The Common Vulnerability Scoring System (CVSS) is a free and open industry standard for assessing the severity of computer system security vulnerabilities.

CWE

Common Weakness Enumeration (CWE™) is a community-developed list of common software and hardware weakness types that have security ramifications. Weaknesses are flaws, faults, bugs, vulnerabilities, or other errors in software or hardware implementation, code, design, or architecture. If left unaddressed, weaknesses could result in systems, networks, or hardware being vulnerable to attack. The CWE List and associated classification taxonomy serve as a language that you can use to identify and describe these weaknesses in terms of CWEs.

Deduplication

When a category’s process deems findings to be the same, or if they are similar enough that a noise reduction is required, only one finding is kept and the others are eliminated. Read more about the deduplication process.

Dependency graph export

A dependency graph export lists the direct and indirect dependencies used by a project and includes the relationships between them. It’s differentiated from a lock file because it may not be required by a package manager during installation like in the case of a pipdeptree graph export.

Duplicate finding

A legitimate finding that is reported multiple times. This can occur when different scanners discover the same finding, or when a single scan inadvertently reports the same finding more than once.

False positive

A finding that doesn’t exist but is incorrectly reported as existing.

Finding

An asset that has the potential to be vulnerable, identified in a project by an analyzer. Assets include but are not restricted to source code, binary packages, containers, dependencies, networks, applications, and infrastructure.

Findings are all potential vulnerability items scanners identify in MRs/feature branches. Only after merging to default does a finding become a vulnerability.

You can interact with vulnerability findings in two ways.

1. You can open an issue or merge request for the vulnerability finding.
2. You can dismiss the vulnerability finding. Dismissing the finding hides it from the default views.

Grouping

A flexible and non-destructive way to visually organize vulnerabilities in groups when there are multiple findings that are likely related but do not qualify for deduplication. For example, you can include findings that should be evaluated together, would be fixed by the same action, or come from the same source.

Insignificant finding

A legitimate finding that a particular customer doesn’t care about.

Known affected component

A component that matches the requirements for a vulnerability to be exploitable. For example, packageA@1.0.3 matches the name, package type, and one of the affected versions or version ranges of FAKECVE-2023-0001.

Location fingerprint

A finding’s location fingerprint is a text value that’s unique for each location on the attack surface. Each security product defines this according to its type of attack surface. For example, SAST incorporates file path and line number.

Lock file

A lock file exhaustively lists both the direct and indirect dependencies of an application to ensure reproducible builds by a package manager. It may also be a dependency graph export like in the case of a Gemfile.lock file, but listing dependency relationships is not a requirement or guaranteed.

Package managers and package types

Package managers

A package manager is a system that manages your project dependencies.

The package manager provides a method to install new dependencies (also referred to as “packages”), manage where packages are stored on your file system, and offer capabilities for you to publish your own packages.

Package types

Each package manager, platform, type, or ecosystem has its own conventions and protocols to identify, locate, and provision software packages.

The following table is a non-exhaustive list of some of the package managers and types referenced in GitLab documentation and software tools.

Pipeline Security tab

A page that displays findings discovered in the associated CI pipeline.

Possibly affected component

A software component that is possibly affected by vulnerability. For example, when scanning a project for known vulnerabilities, components are first evaluated to see if they match the name and package type. During this stage, they’re possibly affected by the vulnerability, and are only known to be affected after it’s confirmed that they fall in the affected version range.

Post-filter

Post-filters help reduce noise in the scanner results and automate manual tasks. You can specify criteria that updates or modifies vulnerability data based on scanner results. For example, you can flag findings as likely false positives and automatically resolve vulnerabilities that are no longer detected. These are not permanent actions and can be changed.

Support for automatically resolving findings is tracked in epic 7478 and support for cheap scan is proposed in epic 7886.

Pre-filter

An irreversible action that is done to filter out targets before analysis occurs. This is usually provided to allow the user to reduce scope and noise and speed up the analysis. This should not be done if a record is needed as GitLab doesn’t store anything related to the skipped/excluded code or assets.

Examples: DS_EXCLUDED_PATHS should Exclude files and directories from the scan based on the paths provided.

Primary identifier

A finding’s primary identifier is a value that is unique to each finding. The external type and external ID of the finding’s first identifier combine to create the value.

An example primary identifier is CVE, which is used for Trivy. The identifier must be stable. Subsequent scans must return the same value for the same finding, even if the location has slightly changed.

Processor

Software that accepts an input and transforms it according to specified criteria, either by modifying the input data or by attaching additional metadata as an output. Processors exist to support scanner operations and are commonly used in pre-scan and post-scan phases. Unlike filters, processors do not have decision-making capabilities to control workflow continuation or termination based on business logic. Instead, they perform transformations and pass the results forward unconditionally.

Pre-processor

Pre-processors typically perform data preparation tasks such as normalizing input formats, enriching scan targets with additional context, applying target-specific transformations, or augmenting configuration parameters. They ensure that the scanner receives properly formatted and enhanced inputs optimized for the scanning operation.

Post-processor

Post-processors apply intelligent analysis to scan results after the scanner completes its operation. Post-processors enhance raw scanner output through operations like vulnerability classification, false positive filtering, severity adjustment, and contextual enrichment. Scanner results can pass through multiple post-processors in sequence before the processed results are returned to the analyzer.

Reachability

Reachability indicates whether a component listed as a dependency in a project is actually used in the codebase.

Report finding

A finding that only exists in a report produced by an analyzer, and is yet to be persisted to the database. The report finding becomes a vulnerability finding once it’s imported into the database.

Scan type (report type)

Describes the type of scan. This must be one of the following:

• api_fuzzing
• container_scanning
• coverage_fuzzing
• dast
• dependency_scanning
• sast
• secret_detection

This list is subject to change as scanners are added.

Scan target type

A discrete unit of content or artifact that serves as the scope boundary for running the scan. Each scan target type represents a self-contained entity with defined scanning constraints. A specific instance of a scan target type (such as a particular Git repository or container image) is referred to as a “scan target”. Examples of scan target types include Git repositories, file systems, containers, etc.

Scanner

Software that scans for security vulnerabilities in a scan target (an instance of a scan target type). It is generally a stateless component that receives necessary scan configuration parameters and scan payloads from the analyzer. The resulting scan report is not necessarily in the Secure report format. A scanner can be a sophisticated component that wraps one or more scan engines with additional processors (for example, the secret detection scanner), or it can be as simple as a standalone scan engine (for example, Trivy).

Secure product

A group of features related to a specific area of application security with first-class support by GitLab.

Products include container scanning, dependency scanning, dynamic application security testing (DAST), secret detection, static application security testing (SAST), and fuzz testing.

Each of these products typically include one or more analyzers.

Secure report format

A standard report format that Secure products comply with when creating JSON reports. The format is described by a JSON schema.

Security dashboard

Provides an overview of all the vulnerabilities for a project, group, or GitLab instance. Vulnerabilities are only created from findings discovered on the project’s default branch.

Seed corpus

The set of test cases given as initial input to the fuzz target. This usually speeds up the fuzz target substantially. This can be either manually created test cases or auto-generated with the fuzz target itself from previous runs.

Vendor

The party maintaining an analyzer. As such, a vendor is responsible for integrating a scanner into GitLab and keeping it compatible as they evolve. A vendor isn’t necessarily the author or maintainer of the scanner, as in the case of using an open core or OSS project as a base solution of an offering. For scanners included as part of a GitLab distribution or GitLab subscription, the vendor is listed as GitLab.

Vulnerability

A flaw that has a negative impact on the security of its environment. Vulnerabilities describe the error or weakness, and don’t describe where the error is located (see finding).

Each vulnerability maps to a unique finding.

Vulnerabilities exist in the default branch. Findings (see finding) are all potential vulnerability items scanners identify in MRs/feature branches. Only after merging to default does a finding become a vulnerability.

Vulnerability finding

When a report finding is stored to the database, it becomes a vulnerability finding.

Vulnerability tracking

Deals with the responsibility of matching findings across scans so that a finding’s lifecycle can be understood. Engineers and security teams use this information to decide whether to merge code changes, and to see unresolved findings and when they were introduced.

Vulnerabilities are tracked by comparing the location fingerprint, primary identifier, and report type.

Vulnerability occurrence

Deprecated, see finding.