SkillData

arbor

Organise repeatable AI research experiments.

MIT licence

What it does

Uses a hypothesis tree to track candidate improvements to models, agent systems or benchmarks. It records isolated experiments and compares promising candidates with held-out evaluation evidence.

Installation

This command adds the whole Scientific Agent Skills collection, which includes arbor.

npx skills add K-Dense-AI/scientific-agent-skills

  • The installer offers the upstream collection, including specialist science skills not listed here. Select only the skills you need; installing instructions does not install their runtime dependencies.
  • Individual skills have their own licence metadata, which can differ from the repository MIT licence.

Installation options in the README

Using this skill

  • Requires Python 3.10+ for the bundled state manager and Git for experiment worktrees. The optional arbor-agent CLI needs a separate installation; autonomous model calls need provider credentials and network access.

Part of a collection