How to use the Nextflow agent skill: install and run pipelines

What the Nextflow skill makes the agent do

The skill is a set of instructions and reference documents for an agent. Once it is installed, the agent can help you run existing Nextflow or nf-core pipelines, write new DSL2 pipeline code, set up configuration for containers and schedulers such as SLURM or cloud batch services, write and run nf-test tests, and diagnose failed or resumed runs.

The skill does not do the scientific or computing work by itself. The actual work happens when Nextflow tools run on your own machine or server, so the agent is guiding and running commands in your environment.

• Run nf-core or custom Nextflow pipelines
• Write or edit .nf files, nextflow.config and profiles
• Build modules and subworkflows with nf-test tests
• Configure containers, executors and resources for HPC or cloud
• Investigate failed runs and runs that resume from cache

How do you install the Nextflow skill?

Download the skill ZIP from this page and unzip it. For a personal setup, place the unzipped skill folder in ~/.claude/skills/. For a single project, place it in that project's .claude/skills/ folder instead, so it applies only there.

In the Claude apps, you can instead upload the ZIP under Settings, then Capabilities, then Skills.

Nextflow itself is a separate install. The skill's Setup section describes a self-installing launcher. Download the installer, review it before running it, then run it and check the result with nextflow info.

• Personal install: ~/.claude/skills/
• Project install: .claude/skills/ inside your project
• Claude apps: Settings, then Capabilities, then Skills, then upload the ZIP

What do you need before using the skill?

The skill needs a terminal and several tools. Nextflow requires Bash and Java in the 17 to 26 range. Verify Java with java -version, because a launcher on your PATH does not prove that a Java runtime is installed.

nf-core tools needs Python 3.10 or newer, but only for nf-core commands. nf-test is needed for testing. Pipelines typically use Docker, Singularity/Apptainer or conda, depending on the pipeline. HPC or cloud deployment also needs scheduler access or cloud credentials.

• Bash 3.2 or newer and Java 17 to 26 (required)
• Python 3.10 or newer (optional, for nf-core commands)
• nf-test (for testing)
• Docker, Singularity/Apptainer or conda, depending on the pipeline
• Scheduler access such as SLURM, or cloud credentials, if you deploy there

How do you trigger the skill with a natural request?

You do not need special syntax. Describe the task in plain words and mention Nextflow, nf-core or a file type such as .nf or nextflow.config. The skill's description lists these trigger terms, so the agent should load its guidance when you use them.

Be specific about your goal. Say whether you want to run a pipeline, write code, configure a cluster or debug a run, and include the relevant files or error messages. The skill separates these tasks into reference documents, so a clear goal helps the agent pick the right one.

• Run the nf-core RNA-seq pipeline on my samplesheet.csv with Docker
• Write a Nextflow process that sorts a BAM file and publishes the output
• Set up nextflow.config to submit jobs to SLURM with container support
• My Nextflow run failed at a process; help me find out why and whether -resume will work
• Write nf-test tests for this module

Safety notes: scripts and network access

The skill runs scripts and uses the network. Its commands can fetch software, containers and data from the internet, and the installer downloads code. Review what the agent plans to run before you approve it.

If you work with sensitive data, review container images, downloads and credentials first. Do not paste secrets into prompts, and check which scheduler or cloud account a configuration would use before submitting a job.

• Review the installer and any commands before running them
• Check container images and downloaded data sources
• Keep credentials out of prompts and config files you share
• Confirm which cluster or cloud account a job will use

Tips for a first successful run

Start with the small test profile of an nf-core pipeline before using real data. A passing test shows that your configuration and the test fixture work. It does not prove scientific accuracy or that your full-scale data will fit on your machine.

Pin versions for reproducibility. Use the -r option to pin a pipeline release, and set the Nextflow version as described in the skill's Setup section. Use -resume to reuse cached results after a fix, but remember that resume is a computational cache, not scientific validation.

• Use the test profile first and check its resource and download needs
• Pin the pipeline revision with -r
• Add -resume when rerunning after a fix
• Use the input schema of the pipeline release you chose