agentsclimarketplace
Sign in

Nextflow debugging

Skill Rebell-Leader/SpatialAI_MCP/.codex/skills/nextflow-debugging

Diagnose a failing Nextflow pipeline run in an OpenProblems benchmark. Use when a `nextflow run` fails, a process errors out, or the user shares a .nextflow.log / .command.err and asks why it broke.From its SKILL.md

Install
npx -y skills add Rebell-Leader/SpatialAI_MCP --skill nextflow-debugging

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 4 stars4 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

2.1 KB, as published. Nobody here has run it

Nextflow debugging

The server does not yet parse logs for you (analyze_nextflow_log is roadmap), so triage the logs directly. Be systematic.

Steps

  1. Find the failing task. From the run output, note the process name and the work directory hash (work/ab/cdef...). The per-task files there are the ground truth:

    • .command.sh — the exact command run.
    • .command.err / .command.out — stderr/stdout.
    • .command.log, .exitcode — exit status.

  2. Classify the error by exit code / message:

    • 137 / OOM-killed → memory limit. Increase the process memory directive or add a dynamic retry (memory { 8.GB * task.attempt }).
    • 127 / "command not found" → tool missing in the container; fix the component's engine/setup or the image.
    • 126 / permission denied → executable bit / entrypoint issue.
    • File/path errors → check channel wiring and input staging; confirm the input file actually exists and matches the expected format (run validate_spatial_data).
    • Config/DSL errors → confirm DSL2 syntax; OpenProblems pipelines are DSL2.

  3. Add resilience where appropriate: errorStrategy 'retry', maxRetries, and dynamic resources by task.attempt. Don't paper over a real bug with retries.

  4. Reproduce in isolation. Re-run a single component with viash run on the test data before re-running the whole pipeline. Use -resume to avoid recomputing successful tasks.

  5. Validate inputs with validate_spatial_data / analyze_spatial_metadata when the failure looks data-shaped (wrong format, missing elements, raw-vs-normalized mismatch).

Output

Report: the failing process, the root-cause class, the evidence (the log line), and the specific fix — plus whether it's a data problem, an environment problem, or a code problem.

Keep looking

Skills are one crate of 627,801. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.