project devlogs / docbank
docbank Own your document archive, with indexed retrieval, stable identities, verifiable content, and recovery for you and your agents
GitHub ↗ Website ↗
By week By project Choose project docbank ↓ AgentsView msgvault roborev kata Kenn Forge Ghosthub docbank kwt BenchDB Fotobank docbank source code growth Source code measured at 17 snapshots, from 0 lines on 2026-07-06 to 656,679 lines on 2026-10-21. Full measurements are in the table below. 0 328,340 656,679 2026-07-06: 0 source lines 2026-07-12: 11,201 source lines 2026-07-19: 55,249 source lines 2026-07-26: 80,563 source lines 2026-08-02: 110,490 source lines 2026-08-09: 110,490 source lines 2026-08-16: 110,490 source lines 2026-08-23: 128,972 source lines 2026-08-30: 169,893 source lines 2026-09-06: 215,068 source lines 2026-09-13: 283,852 source lines 2026-09-20: 417,367 source lines 2026-09-27: 488,620 source lines 2026-10-04: 509,682 source lines 2026-10-11: 540,064 source lines 2026-10-18: 641,525 source lines 2026-10-21: 656,679 source lines 2026-07-06 2026-10-21 docbank source code growth Source code measured at 17 snapshots, from 0 lines on 2026-07-06 to 656,679 lines on 2026-10-21. Full measurements are in the table below. 0 328.3K 656.7K 2026-07-06: 0 source lines 2026-07-12: 11,201 source lines 2026-07-19: 55,249 source lines 2026-07-26: 80,563 source lines 2026-08-02: 110,490 source lines 2026-08-09: 110,490 source lines 2026-08-16: 110,490 source lines 2026-08-23: 128,972 source lines 2026-08-30: 169,893 source lines 2026-09-06: 215,068 source lines 2026-09-13: 283,852 source lines 2026-09-20: 417,367 source lines 2026-09-27: 488,620 source lines 2026-10-04: 509,682 source lines 2026-10-11: 540,064 source lines 2026-10-18: 641,525 source lines 2026-10-21: 656,679 source lines 2026-07-06 2026-10-21 +656,679 lines since 2026-07-06
Measurements & method SCC code lines (nonblank, noncomment) in tracked source files, including tests, SQL, HTML, and CSS. Excludes prose, shell scripts, data/config files, vendored dependencies, build output, generated and minified files. Counts the checked-out tree, not accumulated additions.
Snapshots follow the repository’s default branch. Lines measure codebase size, not work completed. Historical counts can fall when code is removed.
Measured with scc 3.7.0. Updated 2026-10-21.
Source line measurements for docbank Date Source lines Files Commit 2026-10-21 656,679 2,515 77edc97 2026-10-18 641,525 2,478 efb4954 2026-10-11 540,064 2,198 05d3276 2026-10-04 509,682 2,050 eba64ff 2026-09-27 488,620 1,926 7bf6a59 2026-09-20 417,367 1,626 90b7437 2026-09-13 283,852 993 5265c0e 2026-09-06 215,068 727 67ff0ea 2026-08-30 169,893 602 5bf21bb 2026-08-23 128,972 541 f49b1b9 2026-08-16 110,490 446 5551889 2026-08-09 110,490 446 0543318 2026-08-02 110,490 446 0543318 2026-07-26 80,563 364 342d50d 2026-07-19 55,249 266 46e60b5 2026-07-12 11,201 100 7f0147c 2026-07-06 0 0 d8e6fa4
Project timeline Newest first 621 merged PRs · 20 releases · 14 weeks in the devlog
2026-10-12 to 2026-10-18
99 PRs merged (93 Kenn, 6 community) Weekly diff: +128,879 / −24,578 lines
Find photos by person, edit photo metadata, and save smart sets or freeze their results as albums. Compare photos in cull mode and preview rejects before moving them to trash. Export photos as JPEG or PNG in verified ZIP files. Add photos from a folder in the web app or push local folders to a remote daemon with the CLI. Folder uploads and bulk imports are faster. Import M4A recordings and stream WAV or MP3 sources up to 4 GiB. Search content and large folders faster. Duplicate pages load faster, and populated vaults take less time to restart. 2026-10-05 to 2026-10-11
72 PRs merged (66 Kenn, 6 community) Weekly diff: +49,676 / −7,474 lines
Browse photos by timeline or album, inspect them in a loupe, hide them behind a passcode, and recover deleted groups from trash. Large vaults load document catalogs, collections, duplicate sets, content references, and nested paths faster. Pause background jobs across restarts, limit how many run at once, and track them in the activity center. Search can target exact transcript sources and reopen exact quotations from retained text. Docbank can run on authenticated private container networks. 2026-09-28 to 2026-10-04
61 PRs merged (59 Kenn, 2 community) Weekly diff: +41,183 / −15,822 lines
Export sealed snapshots as verified load-file packages, or export selected original files through the CLI and MCP. Import camera groups, search indexed photo metadata, and use smaller photo previews. Create, review, download, inspect, and recover frozen search reports for selected document versions. Imports, large directory listings, searches, storage reports, and tag pages now do less repeated work. Backups and restores now support objects and recovery files larger than 4 GiB, and uploads no longer time out during storage work. 2026-09-21 to 2026-09-27
38 PRs merged (20 Kenn, 18 community) Weekly diff: +90,608 / −7,815 lines
Import received packages and publish stamped PDFs containing only the selected pages. Search files in the web app with natural-language controls, then revisit dated exports from the recent export history. Redacted PDF productions now rebuild from the exact reviewed sources, freeze reviewed inputs, and require approval before export. Email exports now bundle retained message PDFs with their attachment occurrences, and the web inspector safely opens archived email. Docbank now processes EPUB files locally and accepts recoverable PDFs that previously failed format detection. 2026-09-14 to 2026-09-20
59 PRs merged (29 Kenn, 30 community) Weekly diff: +172,895 / −13,420 lines
Import mailbox archives as email documents, retain offline-rendered PDFs, and browse attachments as separate documents. Ingest supported Loom and Cap recordings, process supplied audio, and keep timed transcripts searchable. Export verified document bundles, preview or download them in the web app, and create DAT, CSV, OPT, and Bates-stamped PDF files. Browse saved query results, inspect exact document versions, and navigate verified text and page images. Find documents similar to a stored version, with optional reranking for more relevant search results. 2026-09-07 to 2026-09-13
46 PRs merged (11 Kenn, 35 community) Weekly diff: +95,101 / −2,527 lines
Search now supports saved queries, highlight sets, Boolean and field expressions, duplicate detection, optional expansion and reranking, and explanations for hybrid results. Choose from Gemini multimodal, Cohere, ZeroEntropy, hosted OpenAI, self-hosted BGE-M3, or Qwen3 for embeddings, with reranking support where available. Process documents through the embedded API, authenticated HTTP API, command line, or terminal interface, and view renditions of web documents. Browse and label import collections, recognize tags by color and group, select visible documents, apply tags as one operation, and export selected rows as CSV. Docbank can decode MIME email into verified metadata and index supplied audio using provider transcripts, including transcript-based evidence. 2026-08-31 to 2026-09-06
58 PRs merged (27 Kenn, 31 community) Weekly diff: +58,443 / −8,200 lines
More document formats can now be processed with Docling, PyMuPDF, Marker, Apache Tika, Unstructured, Datalab Convert, LlamaParse, Reducto, Mistral OCR, and Trafilatura. Interrupted jobs can resume without publishing unfinished results. Embeddings now support local OpenAI-compatible services and hosted Voyage or Mistral services. Generations are stored durably, can be activated independently, and use a rebuildable vector index. JPEG, PNG, GIF, WebP, and camera RAW files now get durable, high-density previews. docbank also reads camera and media metadata from more original files, including Canon, Fujifilm, Olympus, Panasonic, and MP4 files. Embedded applications can now create and restore backups, include host files, recover after the original vault is lost, and remove old recovery points to reclaim storage. Local imports can select files with glob patterns, and filtered searches can run without a text query. CSV files can also be converted to PDF within set limits. 2026-08-24 to 2026-08-30
18 PRs merged (1 Kenn, 17 community) Weekly diff: +47,130 / −2,184 lines
Documents can now produce safe Markdown and verified plain-text renditions from the original evidence. Document renditions now have durable records, stable processing fingerprints, and atomic publishing. Backups can include document derivatives without relying on external providers, and cleanup can remove derivatives without deleting originals. Provider uploads now use the exact bytes that passed inspection, and DNS checks apply to the connection actually opened. Preflight now flags cloud placeholders and explains why hydration failed. 2026-08-17 to 2026-08-23
10 PRs merged (10 Kenn, 0 community) Weekly diff: +22,156 / −1,255 lines
Extract text with local OCR without tying document processing to one provider. Normalize documents once and reuse embedding plans across document workflows. Process mixed batches and interleaved documents with media detection and format-specific probing. Mistral document processing and Voyage multimodal embeddings now stop on unsupported or invalid conditions instead of continuing with uncertain results. The v0.13 document understanding APIs now have user documentation. 2026-08-10 to 2026-08-16
2 PRs merged (2 Kenn, 0 community) Weekly diff: +44 / −47 lines
Developers now have clearer document-processing boundaries and simpler imports. Docbank now uses Go 1.26.6 to keep module verification consistent. 2026-07-27 to 2026-08-02
22 PRs merged (22 Kenn, 0 community) Weekly diff: +31,885 / −1,245 lines
Upload verified documents directly from the web app. Assign, inspect, manage, and browse documents by tag on the web. Move documents to recoverable trash and restore them from either the web app or terminal interface. Inspect storage, backups, daemon jobs, and permanent audit evidence more easily. Download retained document versions from the web app and use verified multi-store storage. 2026-07-20 to 2026-07-26
42 PRs merged (42 Kenn, 0 community) Weekly diff: +30,304 / −1,658 lines
Browse vaults in a local web app, including immutable versions, audit history, provenance, and background job health. Download verified documents from the web app or CLI without leaving partial files behind. Search by directory, tag, media type, or modification time, and reuse stable node IDs in CLI commands. Reorganize documents atomically and create virtual directories from the CLI. Embed docbank with bounded document, immutable storage, and provenance APIs. 2026-07-13 to 2026-07-19
68 PRs merged (67 Kenn, 1 community) Weekly diff: +51,931 / −3,624 lines
Search verified plain-text document contents and organize them with stable tags. Edit, replace, inspect, revert, and prune versioned content with integrity checks. Track audited changes including imports, moves, renames, deletions, restores, tag updates, and content revisions. Create, verify, and restore packed backups with audit fidelity checks. Run the full Docbank lifecycle on Windows, with release builds available across six platforms. 2026-07-06 to 2026-07-12
26 PRs merged (26 Kenn, 0 community) Weekly diff: +16,317 / −1,359 lines
Use Docbank's first release to ingest documents, store content, browse a virtual tree, and work through the CLI or daemon. Upload remote content with digest verification and confirm content identity over HTTP. Stream verified content with separate limits for loose and packed objects. Inspect, maintain, and reclaim packed storage through explicit commands. Read documentation routes in Markdown and follow documented workflows for users and agents.