As of version 21.09.0-edge, Nextflow adds the built-in support for pipeline secrets to allow users to handle and manage sensitive information for pipeline execution in a safe manner.


This is a preview feature, therefore options and syntax may change in future release.

How it works

This feature allows decoupling the use secrets in your pipelines from the pipeline code and configuration files. Secrets are instead managed by Nextflow and store separately into a local store only accessible to the secrets owner.

When the pipeline execution is launched Nextflow inject the secrets in pipeline jobs without leaking them into temporary execution files. The secrets are accessible into the job command via environment variables.


This feature needs to be enabled by settings the following environment variable in the launching environment:


Command line

When enabling this feature Nextflow provides a new command named secrets. This command allow four simple operations:




List secrets available in the current store e.g. nextflow secrets list.


Allows retrieving a secret value e.g. nextflow secrets get -n FOO.


Allows creating creating a new secret or overriding an existing one e.g. nextflow secrets put -n FOO -v "Hello world"


Allows deleting an existing secret e.g. nextflow secrets delete -n FOO.

Configuration file

Once create the secrets can be used in the pipeline configuration file as implicit variables using the secrets scope:

aws {
  accessKey = secrets.MY_ACCESS_KEY
  secretKey = secrets.MY_SECRET_KEY

The above above snippet access the secrets MY_ACCESS_KEY and MY_SECRET_KEY previously and assign them to the corresponding AWS credentials settings.


Secrets cannot be assigned to pipeline parameters.

Process secrets

Secrets can be access by pipeline processes by using the secret directive. For example:

process someJob {
    secret 'MY_ACCESS_KEY'
    secret 'MY_SECRET_KEY'

    your_command --access $MY_ACCESS_KEY --secret $MY_SECRET_KEY

The above snippet runs a command in with the variables MY_ACCESS_KEY and MY_SECRET_KEY are injected in the process execution environment holding the values defines in the secret store.


This feature is only available when using the local or batch scheduler executions e.g. Slurm, Grid Engine, etc.