Project Ops

RS AI Learning Lab — Claude Code drafting workbench

Project operations source: RS AI Learning Lab — Claude Code drafting workbench.

Use this portable workbench when Tim prepares a proposed lab for Radar to ingest into the canonical RS AI Learning Lab library. It produces a complete candidate Draft package for review; it does not edit the canonical repository, GitHub Issues, or a project board.

Use it

  1. Copy this folder into an isolated working directory for one lab.

  2. Complete LAB-BRIEF.md with the agreed assignment. Keep exact module and track slugs from the RS project; do not ask Claude to infer them.

  3. Start Claude Code from this folder and run:

    /draft-lab-submission LAB-BRIEF.md
  4. Review delivery/<lab-slug>/ and delivery/HANDOFF.md together. Send both to Radar for ingestion review.

What comes back

The candidate folder has the same shape as a canonical lab package:

delivery/
  HANDOFF.md
  <lab-slug>/
    README.md
    lab.md
    facilitator-guide.md
    example/worked-example.md
    gallery/gallery-entry.md
    materials/                 # only when the brief requires reusable inputs

app/ is prohibited unless the brief carries a separate, explicit decision authorizing a narrow static learning asset.

Ingestion boundary

Radar verifies placement, collisions, relationships, links, source metadata, and review gaps before copying a candidate into projects/rs-ai-learning-lab/labs/. A complete-looking submission is still Draft: it is not approved for learners, a pilot, publishing, or a status advance.

This is a pilot handoff pack for Tim’s contribution lane. It does not replace the project’s Lab Contribution Workflow or establish a broader contributor process.

Claude Code compatibility

The preferred project skill lives at .claude/skills/draft-lab-submission/. Claude Code 2.1.203 or later discovers project skills automatically. This pack also includes a matching .claude/commands/ compatibility shim for older Claude Code installations; it directs Claude to the same skill instructions. Updating Claude Code remains preferable.


Source: contributor-packs/claude-code-lab-drafter/README.md