GitHub Actions
Last updated: Nov 11, 2026
Fluid Attacks has dedicated GitHub Actions for automated security testing directly in your CI/CD pipeline.
| Action | Scanner | Description |
|---|---|---|
| SAST | Static Application Security Testing | Detects vulnerabilities in your source code |
| SCA | Software Composition Analysis | Detects known CVEs in your dependencies |
| Secret Scan | Secrets Detection | Detects hardcoded credentials and secrets |
| DAST | Dynamic Application Security Testing | Detects vulnerabilities in live web applications |
| CI Gate | Platform integration | Blocks merges when policy-breaking vulnerabilities are present |
How it works
The SAST, SCA, and Secret Scan actions have the same main workflow structure. It has a checkout step, the scanner action, and an optional SARIF upload to GitHub's Security tab.
By default, the SAST, SCA, and Secret Scan actions run a full scan, for all triggers.
To scan only the changed files, set the scanner_mode: diff action input.
Fluid Attacks recommends it for SAST.
In differential mode, the base for the comparison depends on the Git event that started the workflow:
- Pull request → changed files relative to the PR base branch.
- Push to the default branch → changed files since the commit before the push.
- Push to any other branch → changed files relative to the default branch.
- Any other event (scheduled or manual) → full scan.
The differential mode keeps CI fast and makes sure that the action finds new vulnerabilities
before they get to the default branch.
It needs fetch-depth: 0 in the checkout step.
If there is no base commit, or no file changed, the action runs a full scan or skips the scan, respectively.
The DAST action has no differential mode.
Authenticate (optional)
When you authenticate, the Fluid Attacks platform attributes each scan to your Fluid Attacks group with a Group
token. On GitHub Actions, give the action the group input.
Give the workflow the id-token: write permission.
Add a trust for the repository.
The run then gets a short-lived token from OIDC and authenticates without a secret.
As an alternative, keep a long-lived Group token as a secret.
Give it in the api_token input.
The token identifies the group, so you do not need the group input.
If you give a token, the action uses it and not OIDC.
Without a token or OIDC credentials, the scan runs without authentication.
Do not give api_token in a pull_request_target workflow that checks out
the head of the pull request. The code of a fork can then read the secret.
Authentication is optional at this time and does not stop a scan. A future release will make it mandatory. Fluid Attacks recommends that you use it, so that the Fluid Attacks platform attributes your runs to your group.
An authenticated SAST, SCA, or Secret Scan run applies the exclusions and the
severity threshold of your group on the Fluid Attacks platform. The exclusions
replace your local exclude list. For more information, read What
authenticating gets
you.
For the setup, read Authentication and OIDC federation.
What it looks like in GitHub
Job steps
After a push or pull request, the workflow appears in the Actions tab with each step: checkout, scan, and upload.

Scanner logs
When you expand the scan step, you see the scanner output. It shows the files or dependency manifests that the scanner found, the vulnerabilities that it detected, and if the run was a full or differential scan.

Security tab results
After the SARIF file uploads, the findings appear in the Security → Code scanning panel of the repository. Each finding shows the severity, file location, and description of the vulnerability.

Privacy: The actions collect only anonymized programming language metadata for analytics. Source code and scan results never leave your runner. See the EULA for details.
GitHub Advanced Security
To upload scan results to the GitHub Security tab, you must have GitHub Advanced Security. It is available on all public repositories and on private repositories with a GitHub Advanced Security license. On private repositories without that license, the upload step fails.
Always use if: always() in the upload step.
Thus, the step uploads the results each time, also when the scan step exits with a non-zero code (for example, when you set strict: true):
- name: Upload SARIF
if: always()
uses: github/codeql-action/upload-sarif@v4
with:
sarif_file: ${{ steps.scan.outputs.sarif_file }}Common scenarios
Strict mode: block merges with vulnerabilities
By default, the actions report vulnerabilities without failing the job.
Set strict: true as an action input to fail on any vulnerability:
- uses: fluidattacks/sast-action@<version>
with:
strict: trueTo break the build only for vulnerabilities at or above a severity that you set,
use strictness_threshold (low, medium, high, or critical).
It breaks the build by itself, with or without strict.
The scan reports all the vulnerabilities and fails only for those that reach the threshold:
- uses: fluidattacks/sast-action@<version>
with:
strictness_threshold: highYou can set the two values also in your configuration file. These rules apply:
- The action inputs override the configuration file, including
strict: falseover a file that saystrue. If you omit an input, the configuration file decides. - In an authenticated run, the minimum CVSS score of your group on the platform also applies.
If a policy and
strictness_thresholdexist, the stricter of the two applies. Thus, the workflow can only make the gate stricter than the platform, never less strict. - The action ignores an invalid value and shows a warning. Then, the configuration file decides.
Then, in the branch protection settings of your repository, enable Require status checks to pass before merging.
Export results as CSV or SARIF
Use output.file_path and output.format in your configuration file
to write results to a file:
output:
file_path: results.sarif
format: SARIFThe correct values for format are SARIF, CSV, and ALL (generates the two formats at the same time).
The action sets the sarif_file output only when the format is SARIF or ALL.
Troubleshooting
The pipeline fails unexpectedly
If you set strict: true or strictness_threshold (as an action input or in your configuration file),
the pipeline fails when the scanner finds vulnerabilities, or ones at or above the threshold.
Remove them, or raise strictness_threshold, to report findings without a pipeline failure.
The action input wins over the configuration file.
The job fails with "not found in repository"
The path in scan_config_path does not exist in the repository.
Make sure that the path is correct and relative to the repository root.
No results appear in the Security tab
Make sure that the upload step uses if: always().
Thus, it runs also when the scan finds vulnerabilities with strict: true.
Also make sure that your repository has GitHub Advanced Security enabled.
For more information, read GitHub Advanced Security.
Differential scan analyzes all files instead of just changes
Make sure that you set scanner_mode: diff and that the actions/checkout step sets fetch-depth: 0.
Without the full git history, the action cannot find which files changed
relative to the default branch.