Query the catalog

chartcoach gives you access to a versioned visualization guideline catalog from the CLI, Python, JavaScript, or MCP.

For agents

Start with the CLI-served core skill, then retrieve candidate rows as JSONL. Read the exact guideline id before citing it.

uvx chartcoach@latest skills get core
uvx chartcoach@latest catalog query --contains "legend" --limit 5 --format jsonl
uvx chartcoach@latest catalog read add-unlabeled-axis-ticks-to-continuous-color-keys \
  --source-detail minimal \
  --format json
uvx chartcoach@latest catalog cite add-unlabeled-axis-ticks-to-continuous-color-keys \
  --format json

Use the id from a selected JSONL row in later read and cite calls. Query and search rows are candidates. Exact read output is the guideline text that can support a claim.

Human inspection

Use table output when reading rows in a terminal:

uvx chartcoach@latest catalog query --contains "legend" --limit 5 --format table

Next steps

• Getting started: run the first query, read one guideline, and cite its sources.
• Querying: filter by text, labels, sections, SQL, and indexed search with compact CLI, Python, and TypeScript examples.
• Cataloging scheme: inspect stable ids, row shape, manifest, source locators, cache behavior, release metadata, and SQL tables.
• CLI, Python, and JavaScript: use the same catalog from terminal commands, notebooks, services, and apps.
• Agents and MCP: route agent work through skills or expose catalog SQL and search as MCP tools.
• Build a catalog: validate authored entries and publish a catalog bundle.