Overview | chartcoach docs

Use chartcoach catalog tables, labels, sections, sources, and JSONL output as context for advanced queries and downstream tools.

chartcoach exposes the Guideline Catalog as records, labels, sections, source references, SQL tables, table output, JSONL rows, and indexed text. Those pieces give shell commands, notebooks, dashboards, and agents the same reusable context with stable ids and source metadata.

chartcoach is built for composability and interoperability: use table output for reading, JSONL for standard tools, and stable ids for follow-up reads and citations.

Start by checking the current table list:

uvx chartcoach@latest catalog schema --tables --row-counts --format table

Output:

Table	Rows
`guidelines`	781
`sections`	6248
`labels`	209
`guideline_labels`	5752
`references`	262
`guideline_references`	1024
`guideline_sources`	1024

Use --format table while exploring in a terminal. Switch to JSONL with jq when another tool should consume the same query:

uvx chartcoach@latest catalog schema --tables --row-counts --format jsonl |
  jq -s 'map({table: .name, rows})'

Each guide starts with one question, runs one catalog query, and shows a small Markdown table.

Question	Guide
Which guidelines cite sources authored by a researcher?	Source authors
Where does the catalog both recommend and warn against a topic?	Use and avoid
Which labels lean on papers, standards, posts, or reports?	Evidence mix
Which chart families cluster around which reader tasks?	Task matrix
Which failure descriptions match reviewer language?	Failure search