data-and-docs-search

Data & Docs Search

Search beyond source code.

This skill is portable across GitHub Copilot, APM, and Claude Code; keep the references/ folder next to SKILL.md when copying it into a Copilot skill location.

Corpus	Use	Reference
JSON	jq, or gron + rg	jq-yq-gron
YAML / TOML / XML	yq	jq-yq-gron
CSV / Parquet / JSON at scale	duckdb, sqlite-utils	data-files
PDFs / Office docs / archives	rga, pandoc, pdftotext	docs

Decision flow: known JSON path → jq | "grep this JSON" → gron \| rg | YAML/TOML → yq | tabular/analytical → duckdb | inside PDFs/docs → rga.

Best practices

1. gron file.json | rg pattern turns nested JSON into greppable lines — ideal when you don't know the exact path. gron -u reverses it.
2. rga is rg for non-text files; first run builds a cache, later runs are fast. Scope with --rga-adapters and the same globs as rg.
3. For analytical queries over CSV/Parquet, duckdb reads files directly: duckdb -c "SELECT ... FROM 'data.parquet' ...".
4. These tools are optional — run scripts/check-tools.sh and install what's missing before relying on them.
5. If rtk is installed, prefix rtk rg … for compact output; the data tools (jq/yq/gron/duckdb/rga/…) aren't rtk-wrapped — run them directly. See the rtk note in code-search.