Scaling tedious-but-necessary repository maintenance using AI agentic systems.

nf-core/modules contains hundreds of modules, many of which carry accumulated technical debt: missing stub blocks, inconsistent test coverage, outdated conventions, and absent documentation. These are the tasks that human developers rarely want to do β€” but they’re essential for the health of the ecosystem.

This project demonstrates how agentic AI swarms can systematically tackle this maintenance at scale, while keeping a human firmly in the loop for review and submission.

Track Record

Before the hackathon, this approach already delivered results:

  • 48 modules identified as missing stub blocks (#4570)
  • 40 modules stubbed and validated using automated AST analysis and nf-test snapshot verification
  • 10 atomic PRs submitted (#11349–#11358), split by domain category for focused review
  • Every stub follows the canonical echo | gzip > convention (#5409)

Resources

The methodology, tooling, and lessons learned are documented in detail:


Goal

Use agentic AI systems to identify and resolve systematic maintenance gaps across nf-core/modules β€” starting from the proven stub block foundation and expanding into additional cleanup workstreams during the hackathon.


Workstreams

1. Stub Block Completion (Proven)

Continue the #4570 initiative. Identify any remaining modules missing stubs, generate convention-compliant stub blocks, validate with nf-test, and submit atomic PRs.

2. Additional Cleanup Targets (Hackathon Scope)

Potential areas to tackle collaboratively during the event:

  • nf-test coverage gaps β€” Modules with incomplete or missing test snapshots
  • Topic channel migration β€” Updating modules to use modern Nextflow topic channels
  • Linting compliance β€” Resolving persistent nf-core modules lint warnings at scale
  • Documentation standardization β€” Ensuring consistent meta.yml and module documentation

The exact scope for workstreams 2+ will be refined at the hackathon based on community priorities and participant interest.


What Participants Will Do

  1. Pick a workstream β€” Choose from stub completion, test coverage, topic migration, or linting cleanup.
  2. Use the agentic workflow β€” or contribute manually, both are welcome.
  3. Validate locally β€” Run nf-core modules lint and nf-test before submission.
  4. Submit atomic PRs β€” One module or small batch per PR for focused review.

This project is suitable for all experience levels. Each task is self-contained and follows established nf-core contribution patterns.


An AI swarm of agents collaboratively cleaning up a repository
category
components
group leader