v0.3.0

Added

• perceiveDirect(url, opts) — one-call raw-byte retrieval: the response body is the artifact itself (markdown, PDF, screenshot…), metadata parsed from X-* headers into PerceiveDirectResult. No second fetch to a signed URL.
• downloadPerceiveArtifact(operationId, output?) — re-download a stored artifact of an earlier operation.
• onlyMainContent option (server default true) — markdown is main-content-only with site chrome stripped behind a fidelity guard; set false for the full page.
• directDownload option on PerceiveOptions.
• New result fields: statusCode (upstream HTTP status of the rendered page), deductions (named render-quality deductions), optionsEcho (the options the server honoured).

Removed

• markdown_fit output — removed by the API. Use markdown (main-content-only by default) instead.

Compatibility

• Requires the 2026-08 gateway for directDownload/onlyMainContent to take effect; older gateways ignore the flags.

v0.2.0

Additive release. Nothing from 0.1.1 was removed or renamed.

New

• compressImage(file, { targetSizeKb }) for POST /v1/convert/compress-image. PNG, JPEG and WebP only. The output keeps the input format and extension and is never larger than the input. Lossless first, with metadata stripped but the ICC profile and EXIF orientation preserved. targetSizeKb is best effort: an unreachable target returns the smallest file achieved rather than throwing, so check result.fileSize.
• width and height on convertImage, for SVG input only (svg-to-png, svg-to-jpeg, svg-to-webp). Each 1 to 10000. One alone preserves the SVG's aspect ratio. Total output capped at 25 megapixels.
• npx enconvert-sdk upgrade CLI, with --dry-run and version. Detects npm, pnpm, yarn or bun and always prints the command before running it.

Improved

• convertToMarkdown and convertToPdf now pre-validate the file extension locally (22 and 36 accepted extensions), so a bad input throws before any upload.
• Corrected the convertToPdf pdfOptions documentation. Full page geometry is honored for HTML, Markdown, plain text, EPUB, image and SVG input. Office, ODF, iWork, RTF and CSV input plus PDF passthrough accept grayscale only.
• Repository URL fixed from the old conversionapi org to enconvert.

Published manually, so this version has no npm provenance attestation. Future releases publish from CI with provenance.

v0.1.0

Full Changelog: https://github.com/conversionapi/node-sdk/commits/v0.1.0