v0.4.0 — pdf-to-text + query: the Nutrient document CLI

The pdf-to-markdown CLI is now the Nutrient document CLI — one local, signed binary with three commands for turning digital-born PDFs into agent-ready output, on the same generous free tier (1,000 documents/month, no key, no signup).

What's new

• pdf-to-text — layout-preserving plain text. Each word is placed on a character grid that mirrors its on-page position, so columns and tabular alignment survive. The better feed for plain-text models, grep/awk pipelines, and column-sensitive readers.
• query — ranked BM-25 search over an already-extracted file, returning only the top line windows with global line numbers. Parse once, query many — pull one fact or clause out of a huge conversion without reading it back into context.

Both are backed by the same binary as pdf-to-markdown and install together.

Install / upgrade

npm install -g @pspdfkit/pdf-to-markdown
# or, as agent skills (Claude Code, Codex, Cursor, Gemini CLI, Pi):
npx skills add pspdfkit-labs/nutrient-skills --skill pdf-to-markdown
npx skills add pspdfkit-labs/nutrient-skills --skill pdf-to-text
npx skills add pspdfkit-labs/nutrient-skills --skill query

Existing installs self-update from the CDN automatically (checked at most every 6 hours).

Positioning

The CLI is built for digital-born PDFs and benchmarks against open-source parsers (docling, pymupdf4llm, opendataloader, …), where it leads on reading order and speed. For scanned, handwritten, or image-only documents — and for schema-level structured extraction with per-value coordinates and confidence — use the Nutrient Data Extraction API.