nextflow-hello-workflow— agent skill

Purpose

Verify that the repository-local Nextflow runtime works and execute a tiny deterministic workflow end to end, including a Slurm executor path.

When to use

• You need a small, local Nextflow smoke test.
• You want a known-good example before adapting a larger workflow.
• You want to confirm that Nextflow can submit tiny jobs to the local Slurm cluster.

When not to use

• You need container orchestration or production pipeline packaging.

Inputs

• examples/main.nf
• Optional --out-dir, --work-dir, and executor settings
• For cluster runs: --executor slurm plus an optional partition

Outputs

• Published greeting files in the chosen output directory
• JSON summary with the runtime version, produced files, and optional Slurm trace/accounting details

Requirements

• The repo-local toolchain at slurm/envs/nextflow-tools
• Local filesystem write access for the work and output directories
• For cluster verification: sbatch and sacct access on the local Slurm cluster

Procedure

1. Run python3 skills/reproducible-workflows/nextflow-hello-workflow/scripts/run_nextflow_hello.py --out-dir scratch/nextflow-hello/results for a local smoke path.
2. Run python3 skills/reproducible-workflows/nextflow-hello-workflow/scripts/run_nextflow_hello.py --executor slurm --partition cpu --out-dir scratch/nextflow-hello-slurm/results --work-dir scratch/nextflow-hello-slurm/work --summary-out scratch/nextflow-hello-slurm/summary.json for a cluster-backed smoke path.
3. Inspect the returned files list and the output directory contents.
4. For Slurm runs, inspect trace_rows and slurm_jobs in the summary.
5. Reuse the pattern in examples/main.nf when bootstrapping larger pipelines.

Validation

• nextflow info succeeds through the repo-local wrapper environment.
• nextflow run completes without errors.
• Four greeting files are published to the output directory.
• For Slurm runs, the trace file contains native job IDs and sacct reports COMPLETED with ExitCode 0:0.

Failure modes and fixes

• java: command not found: run through the provided wrapper script instead of calling nextflow directly.
• Write-path issues: switch --out-dir and --work-dir to writable locations.
• Raw nextflow run treats a relative script name like a remote pipeline: use the wrapper or pass an explicit local path.
• Slurm queue delays: rerun on cpu or adjust the partition if the cluster is busy.

Safety and limits

• This skill verifies local and tiny Slurm-backed executor paths only.
• It does not prove container, cloud, or production-scale execution readiness.

Examples

• python3 .../run_nextflow_hello.py --out-dir scratch/nextflow-hello/results --work-dir scratch/nextflow-hello/work
• python3 .../run_nextflow_hello.py --executor slurm --partition cpu --summary-out scratch/nextflow-hello-slurm/summary.json

Provenance

• Nextflow installation docs: https://www.nextflow.io/docs/latest/install.html
• Nextflow first-script tutorial: https://www.nextflow.io/docs/latest/your-first-script.html
• Nextflow executor docs: https://www.nextflow.io/docs/latest/executor.html
• Slurm quick start: https://slurm.schedmd.com/quickstart.html

Related skills

• snakemake-toy-workflow-starter
• nf-core-pipeline-list
• slurm-job-debug-template