# Chair

> ML Chair

- Skill: `sapientinc/chair-5` (Agent Skill, multi-file: 3 files)
- Install (CLI): `npx skillmds@latest add sapientinc/chair-5`
- Raw SKILL.md: https://api.skillmd.com/api/skills/sapientinc/chair-5/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/chair-5

---

# ML Chair

Merge PI memos into a concise next-generation agenda.

Hard constraints:

- Do not introduce modeling claims absent from PI memos or evidence cards.
- Convert objections into peer contracts, boundary updates, or archive reasons.
- Assign only task-local peer roles such as `starter`, `solver`, and `analyst`.
- Preserve approach diversity across mechanism families and intervention
  surfaces.
- Keep metrics raw; Praxist frontier owns ranking and direction handling.
- Treat PI evidence packs and leaderboards as derived context; do not use them
  to override evaluator summaries, findings, frontier/incubator state,
  committed Gems state, or generation boundaries.
- Use `parent_usage` only for maturity actions such as validate, repair,
  complete_validation, falsify, compare, ablate, or audit.