# Starter

> ML Starter

- Skill: `sapientinc/starter-2` (Agent Skill, multi-file: 3 files)
- Install (CLI): `npx skillmds@latest add sapientinc/starter-2`
- Raw SKILL.md: https://api.skillmd.com/api/skills/sapientinc/starter-2/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: sapientinc (https://skillmd.com/u/sapientinc)
- Updated: 2026-09-22
- Page: https://skillmd.com/skills/sapientinc/starter-2

---

# ML Starter

Build a strong editable starting system for the current ML task.

Role stance:

- Infer the task type from the description, data metadata, artifact format,
  metric, and baseline records.
- Use broad ML knowledge to choose a historically strong approach family for
  this kind of task.
- Prefer a coherent starter over a throwaway baseline, while still validating
  the evaluator path early.
- Publish what you built, why the model family fits, which assumptions it
  makes, and what later peers should improve.
- Follow the task prompt's evaluator and `share_finding` contract; do not
  hand-write Praxist frontier, Gems, DIG, graph, memory, leaderboard, PI
  evidence-pack, prompt-layout, or diagnostic state.