workspace-course-topics

Ask an AI to assess REMARK compliance (baseline tier)

Timing: This assignment is introduced in Class 9 and is due before Class 10.

Prerequisites: You have a public GitHub repository for your course project (REMARK-style), as in Paper topic choice and REMARK repo setup. Completing Open a pull request to add your project to the REMARK catalog first is ideal so you already understand the catalog workflow, but you can run this exercise even if your catalog PR is still open or in revision.

Goal

Use a capable AI assistant to study the official REMARK documentation, understand how levels of REMARK are described there, and obtain an honest check of whether your repo plausibly meets the entry / baseline level—then get concrete, ordered steps to close any gaps for that level.

Upstream, the econ-ark/REMARK README distinguishes at least:

A standard (basic) REMARK — reproducible project, reproduce.sh, CITATION.cff, binder/environment.yml, submission via pull request to the catalog.
A published REMARK — satisfies the standard plus archival steps such as a Zenodo DOI (see ZENODO-GUIDE.md).

For this assignment, only assess against the baseline / standard REMARK (not the published-with-DOI track), unless your instructor says otherwise.

What you will point the AI at

Give the AI read access to the REMARK repo and key docs, for example (paste links or use the AI’s web/repo tools if available):

Repository root: github.com/econ-ark/REMARK
README.md — overview and quick start
STANDARD.md — requirements and specifications
How-To-Make-A-REMARK.md — author-oriented guide
Optional: WORKFLOW.md and .ai-context.md if the model can use them

Also give the AI your project context: your repo URL, and (if helpful) the paths to reproduce.sh, CITATION.cff, binder/environment.yml, and REMARK.md if present.

Prompting strategy (minimum)

Ask the AI to summarize the different REMARK “levels” or tracks described in the REMARK README and related docs, in its own words.
Ask whether your repository, as described (and with any files you paste or that it fetches), appears to satisfy the baseline / standard REMARK expectations—not the published-with-DOI tier.
Ask for a numbered list of concrete actions (files to add, scripts to fix, metadata to fill) that would bring your repo up to that baseline, prioritized with quick wins first.
Critically review the AI’s answer: spot-check against STANDARD.md; flag anything that sounds vague or wrong and follow up with a second prompt.

Use a strong model if you have one (same spirit as Bellman revision and next steps).

Deliverable

Submit (per course channel—Canvas, email, or as directed):

The prompt(s) you used (or a short pasted transcript of the conversation).
The AI’s summary of REMARK tiers / baseline vs published (a few sentences is enough).
The AI’s compliance judgment for your repo against the baseline tier (yes / no / partially, with reasons).
Your prioritized action list — either the AI’s list as edited by you after you’ve removed hallucinations and added anything it missed, or a short paragraph explaining what you will do next.

You do not have to finish every remediation step before Class 10; you do have to show you understood the gap and have a credible plan.