Latest Release: v2.1.0 (Stable)
Release History
High-level release notes for Scrubkit — offline text + metadata extraction for .NET.
v2.1.0
STABLEMulti-Targeting Expansion & CLI Out-of-the-Box Extractor Integration
- Explicit .NET 9.0 Targeting: Added
net9.0to target frameworks across all 14 packages, tool, samples, and test suites alongsidenet8.0,net10.0, andnetstandard2.0. - Scrubkit.Tool Out-of-the-Box PDFs & Images: Integrated
Scrubkit.PdfandScrubkit.Imagesdirectly into the global CLI soscrubkit scanprocesses PDFs and image EXIF metadata by default. - Unified 100% Test Suite Pass: All 575+ tests executing and passing across .NET 8, .NET 9, and .NET 10 runtimes.
v2.0.0
STABLE
campaign Read Public Announcement →
2026-10-02
Modular Zero-Dependency Core Split & Fluent Extractor Registry
- Zero-Dependency Core
Scrubkit: Core engine features built-in BCL extractors for Word (.docx), Excel (.xlsx), PowerPoint (.pptx), HTML (.html), RTF (.rtf), and Plain Text (.txt) with zero third-party dependencies. - New Package
Scrubkit.Pdf: Extracted PDF document parser into a modular package powered byPdfPigwith fluentoptions.AddPdf()registration. - New Package
Scrubkit.Images: Extracted image EXIF parser into a modular package powered byMetadataExtractorwith fluentoptions.AddImages()registration. - Unified
.AddAllExtractors(): TheScrubkit.Allmeta-package bundles all 14 packages and provides a one-lineoptions.AddAllExtractors()method to auto-register all formats. - 575+ Test Suite: Added comprehensive theory test suites covering concurrency, boundaries, and escaping across .NET 8 and .NET 10.
- Dependency Modernization: Upgraded
PdfPigto 0.1.16,MinVerto 8.0.0,Microsoft.SourceLink.GitHubto 10.0.401,Microsoft.NET.Test.Sdkto 18.10.1, andxunit.runner.visualstudioto 4.0.0.
v1.12.0
STABLEEngine Hardening, Micro-Optimizations & RAG Recipes
- Zero-Allocation Text Normalization: Optimized whitespace scanning with
Span<T>to achieve zero string allocations for clean inputs. - ReDoS Resilience: Enforced explicit 1-second execution timeouts across all built-in regex redaction rules.
- RAG Ingestion Sample: Published
Scrubkit.Sample.RagIngestionand developer guides forMicrosoft.Extensions.AI,SemanticKernel, vector databases, and Parquet.
v1.11.0
STABLEPlatform Expansion — .NET 10.0 (LTS) Support
- .NET 10.0 Multi-Targeting: Added official
net10.0framework support across all 12 Scrubkit packages, CLI tools, and extensions. - Broad Framework Compatibility: Multi-targets
netstandard2.0,net8.0, andnet10.0.
v1.10.0
STABLEAI Ecosystem Integrations
- Microsoft.Extensions.AI: Added
Scrubkit.Extensions.MicrosoftExtensionsAIfor local PII/secret redaction middleware (IChatClientandIEmbeddingGenerator). - Semantic Kernel Extensions: Added
Scrubkit.Extensions.SemanticKernelfor direct memory and vector store ingestion.
v1.9.0
STABLEBinary Legacy Office Support (.doc, .xls, .ppt)
- Legacy Office Formats: Introduced
Scrubkit.LegacyOfficeproviding zero-dependency text & metadata extraction for pre-2007 Word (.doc), Excel (.xls), and PowerPoint (.ppt) files.
v1.8.0
STABLEHTML/RTF Clean Text, Outlook .msg & Masking
- Clean Text Extractors: Strips markup and extracts metadata from HTML and RTF documents.
- Outlook .msg Support: Added binary
.msgemail extraction toScrubkit.Email. - De-identification Options: Supported stable tokens, salted token hashing, and format-preserving last-digit masking.
v1.7.0
STABLECustom Redaction Rules & Rules JSON
- Caller-Defined Rules: Configure custom regex patterns with custom tokens and timeout safety guards.
- CLI Rules Configuration: Load custom redaction rules and allow/deny lists via
scrubkit --rules <file.json>.
v1.6.0
STABLEIncremental Scans & Sidecar Manifests
- Incremental Processing: Rescan folders by skipping unchanged files using lightweight
Manifestbaselines. - Path Exclusion: Added
ReadOptions.ExcludePathsto prevent self-ingestion loops during CLI or pipeline scans.
v1.5.0
STABLEGlobal CLI Tool, Secret Detection & RAG Chunker
- Global CLI Tool (
Scrubkit.Tool): Command-line folder scanner supporting zero-code extractions to CSV, JSON, JSON Lines, and Parquet. - Secret & Credential Scrubbing: Standard redaction for PEM private keys, JWTs, API keys, and connection strings.
- RAG Text Chunker: Whitespace-snapped overlapping window text chunker for vector embeddings.
v1.4.0
STABLETable Serialization, Parquet Export & Meta-Package
- Table Serialization: Built-in zero-dependency CSV and JSON table writers.
- Parquet Support (
Scrubkit.Parquet): Columnar Parquet table exporter for large-scale dataset storage. - Meta-Bundle (
Scrubkit.All): Single package installation referencing the complete Scrubkit library family.
v1.3.0
STABLEEPUB Extractor & Dependency Injection
- eBook Extractor (
Scrubkit.Epub): Reads.epubdocuments and extracts book spine text and metadata. - Dependency Injection (
Scrubkit.Extensions.DependencyInjection):services.AddScrubkit(...)host registration.
v1.2.0
STABLEMIME Email (.eml) & OpenDocument (.odt, .ods, .odp)
- Email Extractor (
Scrubkit.Email): Extracted headers and message bodies from.emlfiles. - OpenDocument Extractor (
Scrubkit.OpenDocument): Zero-dependency text & metadata extraction for OpenOffice / LibreOffice files.
v1.1.0
STABLEOffline Positioning & Opt-in Redaction Architecture
- Opt-In Redaction: Repositioned core engine around offline extraction with opt-in caller-driven redaction.
- Offline Guarantee Tests: Enforced automated unit tests preventing any network API dependencies.
v1.0.0
STABLEInitial Stable Release
- Stable API: Production release under Semantic Versioning guarantees.
- Package Validation: Enabled API surface compatibility checks between .NET 8 and .NET Standard 2.0.
v0.1.0-preview.1
PRE-RELEASEInitial Public Preview
- Core Engine: Initial release of
FolderScrubber, PDF, Office (docx/xlsx/pptx), plain text, and EXIF extraction.