Example: dag-showcase

dag-showcase is the best first project to open in Rime Editor because it is small enough to understand in one sitting and broad enough to show the product.

Project Shape

The example starts with patient demographics and longitudinal lab visits:

- id: patients_source
  kind: source
  path: data/patients.csv

- id: labs_source
  kind: source
  path: data/lab_visits.parquet

It then rolls labs up per patient, joins demographics to lab features with SQL, builds expression-derived risk features, and ends with report-friendly statistical nodes.

Why It Shows The Editor Well

Workflow	Where to look
File ingress	patients_source, labs_source
Expression language	lab_load, risk_index, repeat_visitors
SQL with named inputs	patient_lab_wide, sql_cohort_refine
Dataset scanning	patient_lab_wide, risk_index, site_outcomes
Statistical outputs	crp_vs_baseline, site_age_chisq
Report preview	the generated report tab

Walkthrough

1. Open examples/dag-showcase.
2. Run the DAG.
3. Select patient_lab_wide.
4. Confirm the preview shows a joined patient/lab table and the SQL source.
5. Select lab_load and inspect the derived feature column.
6. Select site_outcomes and check that the aggregate output is one row per site.
7. Open the report preview and inspect the output sizes and statistical sections.

What To Notice

The point of the example is not the medical story. The point is the review loop:

• graph structure tells you where data came from
• table previews show what each step produced
• YAML/spec remains available when visual controls are not enough
• statistical nodes produce reportable object outputs
• the report collects the same run evidence into a shareable artifact

Expression Nodes In The Demo

The feature-building nodes are deliberately readable:

- id: risk_index
  kind: derive
  inputs: [lab_load]
  as: risk_index
  expr: "coalesce([crp_mean], 0) * 2.0 + coalesce([ldl_max], 0) * 0.05"

This is a good dividing line for Rime Editor: if a formula is readable in the expression language, keep it as a core node so reviewers can inspect it directly. If it becomes multi-step logic, promote it to SQL/Python/R/JavaScript.

Report Output

The report should make the pipeline reviewable without reopening the editor:

• DAG overview with output sizes
• table sections for data-producing nodes
• stat blocks for object-producing nodes
• warnings near statistical results
• browser-openable HTML artifact