PDF Forge

Turn PDF collections into searchable, diffable, source-traceable Markdown that coding agents can use as repository context — fully local. Documents never leave your machine.

What it does

Core PDF processing is fully local and works offline. Documents never leave your machine. PDF Forge views PDFs with bundled pdf.js (no CDN), extracts text with pdf-parse, and — new in v1.0.9 — batch-converts a folder of PDFs into a Markdown corpus your editor and coding agents can search and diff, with source paths recorded in front matter and the index.

Identity: babyfox1306.pdf-forge (Marketplace and Open VSX).

Convert Folder to Markdown (v1.0.9)

Command: PDF Forge: Convert Folder to Markdown

Recursively finds .pdf files under a workspace folder
Writes a source-traceable tree under <workspace>/pdf-forge-exports/
Maintains .pdf-forge-manifest.json (ownership, hashes, statuses) and INDEX.md
Soft 100-page monthly batch threshold (whole-file accounting; free regenerations when unchanged). Single-file commands stay unlimited.
Safe writes: does not overwrite user-modified or unowned Markdown
Conflict candidate (at most one): document.pdf-forge-new.md
Cancellable; sequential; no Git side effects on the batch path
Fully local — documents never leave your machine

Batch output layout (example for docs/api/authentication.pdf):

pdf-forge-exports/
  INDEX.md
  .pdf-forge-manifest.json
  docs/api/authentication/document.md

Legacy single-file export (unchanged layout):

pdf-forge-exports/<basename>/<basename>.md

PDF Viewer

Custom editor with canvas page rendering (pdf.js, bundled offline)
Toolbar: zoom, fit, search, extract, export, Copy w/ Citation, sidebar
Auto-reload when the PDF file changes on disk
Output channel PDF Forge for diagnostics

Text extraction & single-file export

Extract full document text to extracted-text.txt
Export to Markdown with YAML front matter
Optional Git auto-commit of the exact generated file when pdf-forge.autoCommit is enabled (default: false)

Code intelligence, tables, compare, notes

Heuristic code-block detection, dedupe, clipboard copy
Tables → CSV / JSON
Compare two PDFs (text diff)
Notes beside the PDF (.notes.json) + Markdown export

Settings

Setting	Default	Meaning
`pdf-forge.autoCommit`	`false`	Commit only the exact single-file export result. Batch never commits.
`pdf-forge.autoReload`	`true`	Reload PDF when it changes on disk
`pdf-forge.maxFileSize`	`200`	Max PDF size in MB (`0` = unlimited)
`pdf-forge.enableOcr`	`false`	Reserved — OCR is not active

Not in this version

Topic	Reality
OCR / scanned PDFs	Not available (no OCR engine ships; image-only PDFs get `no_text` in batch)
MCP / CLI / CI / watch mode	Not shipped
Payment / Pro / licensing UI	Not shipped
Telemetry / install ID / fingerprints	Not shipped
Hidden network / document upload	Not shipped

Privacy

PDF processing is fully local. PDF Forge does not upload documents or document text and makes no background network requests. If you explicitly choose to open the batch guide, your browser visits a public GitHub Pages URL.

Installation

VS Code Marketplace

Extensions (Ctrl+Shift+X)
Search PDF Forge
Install babyfox1306.pdf-forge

Open VSX (VSCodium and forks)

Install babyfox1306.pdf-forge from Open VSX.

From VSIX

Extensions: Install from VSIX… and select the packaged .vsix.

Commands (selection)

PDF Forge: Convert Folder to Markdown
PDF Forge: Export to Markdown
PDF Forge: Extract All Text
PDF Forge: Open Export Folder
PDF Forge: Compare PDFs
PDF Forge: Show Metadata

License

See LICENSE.