Ctrlk
Start FreeLogin
For the complete documentation index, see llms.txt. This page is also available as Markdown.

Secrets

SonarQube Cloud supports the analysis of Secrets, pieces of user-specific or system-level credentials.

Secrets are pieces of user-specific or system-level credentials that should be protected and accessible to legitimate users only.

Configuring secret-specific parameters (general procedure)

To discover and update the Secret-specific properties, navigate in SonarQube Cloud to Your Project > Administration > General Settings > Languages > Secrets. See Configuration overview for more information about setting properties.

How secrets detection works

Secrets detection is not a generic password scanner. It is a layered pipeline that intentionally filters aggressively to keep false-positive rates low. Each layer below is a potential reason why an apparent credential — for example a hardcoded password = "mypassword" in a README.md — does not produce an issue.

Layer 1 — Which files are scanned

  • Binary file extensions (~400) are blacklisted. No content is ever read for these files. The list covers archives (.zip, .tar, .7z…), images (.png, .jpg…), compiled artifacts (.class, .dll, .so…), Office documents (.docx, .xlsx…), and many more. Files whose name looks like an MD5 or SHA hash (32–64 hex characters) are also silently skipped as binary-named cache files.

  • Documentation and template extensions are skipped by every secret rule. This is the most common reason a hardcoded value in a Markdown file is not reported.

    Extension
    Reason

    .md / .mdx

    Markdown documentation

    .html

    HTML pages

    .adoc

    AsciiDoc documentation

    .example / .sample / .template / .dist

    Template or example files not meant to carry real secrets

  • Development environment and translation paths are rejected by every secret rule. These paths typically hold fake or non-sensitive values:

    • .NET development profiles: appsettings.Development*.json, appsettings.Local*.json, launchSettings.json

    • Spring Boot dev profile: application-dev.properties

    • Dev or test .env files: .env.dev*, .env.test*

    • Docker Compose overrides: docker-compose.override.yml, compose.override.yaml, etc.

    • VS Code devcontainer: .devcontainer/devcontainer.json

    • DDEV local dev: .ddev/**

    • i18n / translation files: **/locales/**/*.json, **/i18n/*.json, **/resources/messages*.properties, etc.

  • Only files with an assigned language or matching sonar.text.inclusions are analyzed. When sonar.text.inclusions.activate is true and git integration is available, files outside the inclusion patterns and without an assigned language (for example a plain .txt or a config with an unusual extension) are not scanned unless you extend the inclusions.

  • Automatically detected test files are filtered out. Findings raised in those files are not reported by default. Use sonar.secrets.disableTestFileDetection to surface them as low-confidence instead. This has no effect on files explicitly declared as tests via the sonar.tests property. A file is treated as a test file when any of the following is true:

    • Filename signals: starts with test (e.g. testConfig.properties), contains test. or tests. (e.g. myapp.test.env), or ends with .spec.js, .spec.ts, .spec.jsx, .spec.tsx, _spec.rb, _test.rb, .t.

    • Directory signals (any segment in the path): named test, tests, e2e, mock, mocks, __tests__, __fixtures__, testdata, fixtures, doc, docs, or any segment ending in test or tests (e.g. backend-test/).

Layer 2 — Entropy filter

After a regex pattern matches a candidate string, the candidate is checked for Shannon entropy, a measure of how random a string looks. Strings with many repeated or predictable characters score low and are silently dropped; only candidates above a minimum threshold are kept.

Example value
Result

password

Filtered out

Password123

Filtered out

hunter2

Filtered out

abc123def

Filtered out

ku71CpsLfn8NYhhforGzCRL0

Passes

sk-proj-aB3dEf7GhI…

Passes

A hardcoded value like password = "mypassword" or api_key = "123456789" is never reported because the matched string has too little entropy. Use sonar.secrets.disableEntropyFilter to surface these as low-confidence findings instead.

Layer 3 — Fake-password filter

Even when entropy is sufficient, regex-based post-filters reject values that look like well-known placeholder patterns. Individual rules add provider-specific entries on top of these shared placeholder patterns.

Why an obvious "password" in a file is not reported

Working through the layers, common reasons include:

  1. The file extension is in the shared rejected-extensions list (e.g. .md) — skipped by every rule. This is the most likely reason.

  2. The value is a simple word (e.g. password, mypassword, 123456) — rejected by the fake-password filter before entropy is even checked.

  3. The value has low entropy (e.g. letmein, abc123def) — rejected by the entropy filter.

  4. The file is in a docs/ or test/ directory — silently skipped by test-file detection.

  5. The file extension is not in the inclusions list — never reached by the scanner at all.

All of these layers are intentional. The goal is to surface real, leaked credentials — not every string that looks vaguely secret-like. To deliberately raise low-confidence findings for tuning or evaluation, see Surfacing low-confidence findings.

Adjusting the secret detection scope

By default, SonarQube Cloud detects exposed secrets in all files processed by the language analyzers. You can refine the scope of the secret detection by:

  • Excluding hidden files from the analysis.

  • Adding files based on path-matching patterns.

  • Adjusting the binary file exclusion setup.

Analysis of hidden files

Depending on which scanner is used, additional hidden files tracked by Git are included in the secrets analysis.

This behavior can be disabled by setting the sonar.scanner.excludeHiddenFiles analysis parameter to true.

Analyzing additional hidden files is currently only partially supported with the SonarScanner for Maven and Gradle. Additional hidden files are only analyzed if they’re inside the src/main/java or src/test/java folder in the root or module directories.

Analyzing additional hidden files is currently not supported with SonarScanner for .NET.

Adding files based on path-matching patterns

If you’re using a git repository, you can add files to the secret detection scope by defining path-matching patterns: the files matching the patterns will be included provided they are tracked by git.

To add additional files to the secret detection:

  1. In the SonarQube Cloud, go to Your Organization > Your Project > Administration > General Settings > Languages > Secrets.

  2. Enable the Activate inclusion of custom file path patterns option.

  3. In the List of file path patterns to include, adjust the default path-matching patterns if necessary. See the Defining matching patterns page for instructions.

Alternatively, configure the parameters listed below on the CI/CD host (see the Analysis parameters page for more information).

Property

Description

sonar.text.inclusions.activate

Enables the inclusion of files to the secret detection according to the path-matching patterns defined in sonar.text.inclusions.

sonar.text.inclusions

Comma-separated list of path-matching patterns.

Possible values: A path can be relative (to the sonar.projectBaseDir property, which is by default the directory from which the analysis was started) or absolute. See also the Defining matching patterns page.

Default value: /.sh,/.bash,/.zsh,/.ksh,/.ps1,/.properties,/.conf,/.pem,/.config,/.env,.aws/config,/.key

The default pattern for .env files was extended from .env (root-only) to /.env (any depth). Files like config/production.env or envs/staging.env are now scanned without any configuration change.

Adjusting the binary file exclusion setup

SonarQube Cloud excludes binary files from the analysis. In case binary file types are still included in your analysis, you can exclude these additional files.

To do so:

  1. In the SonarQube Cloud UI, go to Your Organization > Your Project > Administration > General Settings > Languages > Secrets.

  2. In Additional binary file suffixes, enter the list of suffixes to be excluded.

Alternatively, configure the parameter below on the CI/CD host (see the Analysis parameters page for more information).

Property

Description

sonar.text.excluded.file.suffixes

Comma-separated list of additional binary file suffixes to be excluded.

Surfacing low-confidence findings

For benchmarking, evaluation, or to investigate why an expected secret is not reported, you can disable the entropy filter and the automatic test-file detection. Matches that would normally be silently dropped are then reported as low-confidence issues, with the filter name appended to the issue message.

Configure these properties in Your Organization > Your Project > Administration > General Settings > Languages > Secrets, or pass them on the CI/CD host.

Property

Description

Default

sonar.secrets.disableEntropyFilter

When true, low-entropy matches (e.g. placeholder or example secrets) that would normally be silently dropped are reported as low-confidence issues, with the filter name appended to the issue message. Useful for benchmark or evaluation projects. UI name: Disable the entropy filter for secret detection.

false

sonar.secrets.disableTestFileDetection

When true, files automatically identified as test files are no longer filtered out. Findings in those files are reported but marked as low-confidence, with the filter name appended to the issue message. Has no effect on files explicitly declared as tests via sonar.tests. UI name: Disable automatic test-file detection for secret detection.

false

When either property is enabled, the sensor logs a single INFO line at analysis start listing the skipped filters:

Parallel code scan

By default, the analyzer tries to parallelize the analysis of compilation units; it spawns as many jobs as logical CPUs available on the machine.

If required, it is possible to customize the number of scheduled parallel jobs by configuring the property sonar.text.threads=n at the scanner level, where n is an integer indicating the maximum number of parallel jobs.

You should consider setting the sonar.text.threads property only when the automatic detection of the number of logical CPUs cannot detect the desired number. A typical example is when the analysis should not consume all the available computing resources to leave room for other tasks running in parallel on the same machine.

When setting the sonar.text.threads property, you should set it to a value less or equal to the number of logical CPUs available. Over-committing does not accelerate the analysis and can even slow it down.

Analysis of files that don't contain code

Files that don’t contain code (for example, build.gradle and sonar-project.properties) are scanned durning analysis and displayed in the SonarQube Cloud UI after an issue is detected in them. If no secrets are detected in those files, they are not displayed in the UI.

In-line suppression of issues

Issues raised by standard pattern-based secret rules can be suppressed in code files provided the respective language analyzer supports an in-line suppression mechanism.

In-line suppression of issues is not supported on files with no associated language (.env, .pem, .key, .txt, .conf, plain config files).

For more information about the in-line suppression mechanism supported by a given analyzer, see the corresponding section in Languages.

Deactivating secrets analysis

You can deactivate the analysis of secrets by setting the sonar.text.activate property to false.

PreviousScalaNextShell

Last updated

Was this helpful?