2026-06-24 - 2026-07-01
Overview
4 pull requests merged by 1 user
Merged
#48 chore: sprint skill + self-review gate in the dev loop
Merged
#3 docs(plan): v1 sprint-based implementation plan
Merged
#2 docs(plan): make Windows CI runner wiring in-scope Sprint-0 issues
Merged
#1 docs: research findings and v1 design (IR contract, index design, ADRs)
1 pull request proposed by 1 user
Proposed
#49 chore: repo scaffolding & tooling
44 issues created by 1 user
Opened
#4 S0-1: Repo scaffolding & tooling
Opened
#5 S0-2: Linux CI (ci.yml) — green from commit one
Opened
#6 S0-3: The seven interfaces + engine-agnostic types
Opened
#7 S0-4: IR CanonicalDocument model + schema validation
Opened
#8 S0-5: FileSink + fake ModelBackend
Opened
#9 S0-6: Config loader + SQLite StateStore
Opened
#10 S0-7: MeilisearchSink + index auto-create & settings
Opened
#11 S0-8: FilesystemSource + Router + StubExtractor
Opened
#12 S0-9: CLI + trivial end-to-end slice
Opened
#13 S0-10: [forgejo-stack] Register the Windows CI runner
Opened
#14 S0-11: [forgejo-stack] Vendor Windows host toolchain + docs
Opened
#15 S0-12: Wire the Windows CI job (ci-windows.yml)
Opened
#16 S0-13: Heavy-tier placeholder (ci-heavy.yml)
Opened
#17 S1-1: OCR harness bake-off (EARLY, blocking)
Opened
#18 S1-2: OllamaOCRBackend (ModelBackend)
Opened
#19 S1-3: ScannedOcrExtractor
Opened
#20 S1-4: Router text-layer probe & OCR routing
Opened
#21 S1-5: Chunking Transformer — segment packer
Opened
#22 S1-6: Incremental run + clean replace/delete
Opened
#23 S1-7: Concurrency + model-call queue
Opened
#24 S1-8: Sprint-1 OCR e2e slice + fixtures
Opened
#25 S2-1: DoclingExtractor (native-digital + OOXML)
Opened
#26 S2-2: OfficeMetadataAugmenter (in-extractor)
Opened
#27 S2-3: Chunking Transformer — docling-core HybridChunker path
Opened
#28 S2-4: Legacy/Access → skipped routing
Opened
#29 S2-5: SearchProvider Meilisearch adapter
Opened
#30 S2-6: FastAPI + HTMX search UI (search-as-you-type + facets)
Opened
#31 S2-7: Arabic RTL styling + open/locate file
Opened
#32 S2-8: Admin /status + reindex command
Opened
#33 S2-9: Extend Docker Compose with the ui service
Opened
#34 S3-1: Docling ASR backend (large-v3)
Opened
#35 S3-2: A/V IR mapping (segments + timestamps)
Opened
#36 S3-3: A/V chunking + deep-link player + e2e slice
Opened
#37 V2: Legacy binary Office (.doc/.xls/.ppt)
Opened
#38 V2: Access databases (.mdb/.accdb)
Opened
#39 V2: Dedicated faster-whisper A/V extractor
Opened
#40 V2: Vector / hybrid search
Opened
#41 V2: .msg / ZIP / edge-case XML + MarkItDown fallback
Opened
#42 V2: Watched-folder / scheduled / background-service mode
Opened
#43 V2: UI-driven model selection (admin)
Opened
#44 V2: Heavy real-model CI tier activation
Opened
#45 V2: Native Windows installer (Inno Setup/MSI)
Opened
#46 V2: Stop-words + relevance tuning
Opened
#47 S0-14: Minimal Docker Compose (Meilisearch + pipeline, zero-config first run)