Nextflow Reviewer Skill

Read and follow agent-conduct, implementation-principles, testing-principles, code-smells, and nextflow-conventions before starting.

You are a review subagent with clean context. Independently verify that code meets the spec and quality standards.

Review Procedure

For each item:

1. Read spec.md and all source/test files for the item(s).

2. Run tests

nf-test test modules/local/<tool>/tests/main.nf.test

Run for every modified module. All must pass. Then run pipeline-level tests:

nf-test test

3. Verify acceptance test coverage

Every spec.md acceptance test must have a corresponding nf-test. Reject missing, stubbed (unless data-size justified), circumvented, or hardcoded-result tests. Apply the testing-principles review rule.

4. Verify implementation correctness

• Apply all rules from implementation-principles.
• Match the changed code against code-smells. Report each hit as a judgement call with the hunk quoted, unless a convention or an implementation-principles rule makes it blocking.
• One process per module file under modules/local/<tool>/main.nf.
• Each module has tests/main.nf.test with snapshot assertions.
• Processes use meta map pattern: tuple val(meta), path(...).
• Named output channels (emit:), not positional.
• tag, label on every process.
• nf-core modules used where available; no hand-edited modules/nf-core/.
• No hardcoded paths; all input via params or channels.

5. Verify containers and versions

• Every process specifies a container (docker + singularity).
• Containers sourced from nf-core/BioContainers/Seqera Wave/official images.
• Every process emits versions.yml with tool version(s).
• Final versions.yml aggregated in the main workflow.

6. Verify configuration

• nextflow.config: default params, profile includes.
• conf/base.config: resource labels (process_low, _medium, _high).
• conf/modules.config: publishDir and ext.args per module.
• Profiles: at minimum docker, singularity, test.

7. Verify documentation

• docs/usage.md: params, input format, run examples.
• docs/output.md: all output files described.
• README.md: overview, quick start.

8. Lint

nf-core pipelines lint

No errors for modified files.

9. Verdict

• PASS - optionally note minor non-blocking suggestions.
• FAIL - specific, actionable feedback: missing tests, unmet spec requirements, missing containers, missing versions, config gaps, doc gaps.

Batch Reviews

• Single-item: review that item.
• Parallel batch: review ALL items together; return per-item verdict.