v2.12.0 - PDF Ingestion, AI OCR Fallback & LiteLLM Model Discovery

What's New in v2.12.0

📄 PDF Extraction with PyMuPDF & Multimodal AI OCR Fallback

• High-Performance Digital Extraction: Fast text and layout extraction via PyMuPDF with per-page metadata, table structure preservation, and header/footer normalization.
• Vision AI OCR Fallback: Automatically invokes upstream multimodal models (e.g. Gemini 2.5 Flash, Qwen3-VL, Claude 3.5 Sonnet) for scanned, image-only, or dense technical PDFs (such as ASD-STE100 specifications) with page-level rendering.
• PDF Preview & Inspection Modal: Interactive UI modal in Local Storage allowing operators to inspect page text, review AI OCR extraction quality, view generated semantic chunks, and confirm ingestion before writing vectors to Qdrant/pgvector/Chroma.
• MCP Storage Tool Integration: Extended manage_local_file MCP tool with preview action and formatted multi-page PDF reading.

🧠 Dynamic LiteLLM Model Discovery & Hot-Reloadable AI Configuration

• Live Model Discovery: REST endpoint GET /admin/api/models/discover queries upstream LiteLLM gateways (GET /v1/models) and categorizes models into Dense Embeddings, Vision OCR, and General Chat models.
• UI Model Selectors & Discovery Controls: Replaced static text inputs with dynamic capability dropdowns and custom manual write-in options in the Settings tab.
• Relational Metadata Persistence: Model selections (dense_model, sparse_model, vision_ocr_model, chat_model, litellm_url, litellm_api_key) persist to SQLite/PostgreSQL system_metadata and hot-reload in-process without container restarts.

🩺 Vector Database Health Check Fix

• Directly interrogates the active vector store's .health_check() method instead of relying solely on connection status flags, correctly reflecting operational status in the dashboard.

🧪 Automated Test Verification

• Pytest Backend: 492 passed (100%).
• Vitest Frontend: 286 passed across 27 test files (100%).
• Automated CI/CD: End-to-end GitHub Actions workflow validation.

v2.11.0 - Incremental Ingestion, Chunk Cache & Doc Graph Linking

What's New in v2.11.0

🚀 True Incremental Git Syncing

• File-level SHA256 content hashing (compute_git_repo_delta) detects Added, Modified, Deleted, and Unchanged files.
• Skips re-indexing on empty commit changes (0 file deltas).
• Scopes vector store purges (delete_by_path) and SQLite metadata deletions strictly to Modified and Deleted files without wiping the entire repository index.
• Ingests Added and Modified files in chunked batches of 25 files with full rollback and error resilience.

⚡ Chunk-Level SHA256 Embedding Cache

• Adds embedding_cache table in SQLite to cache chunk hash -> dense & sparse embeddings.
• Bulk cache lookups before calling FastEmbed ONNX or LiteLLM API, ensuring identical code and documentation chunks are never re-embedded across commits, branches, or re-indexes.
• Added MAX_FILE_SIZE_BYTES (500KB) guard to skip oversized or minified bundle artifacts.

🕸️ Markdown & Wikilink Topology Graph Extraction

• Extracts standard relative markdown links [Title](../path/to/doc.md) and Obsidian wikilinks [[Doc Name]] / [[Doc Name|Label]] into ast_relationships with relationship type DOC_LINKS_TO.
• Dynamically generates graph edges between documentation nodes in the visual topology graph.

🧪 Test Coverage

• 502 total automated tests (380 backend, 96 frontend, 26 E2E) passing with 100% success rate.

v2.10.0 - Local Embedding Resource Limits & In-App Dynamic Controls

🚀 What's New in v2.10.0

🧠 Local Embedding Resource Limits & Safe Defaults

• System & Cgroup Resource Detection: Automatically detects host and container resource boundaries (cgroup v2 /sys/fs/cgroup/cpu.max, cgroup v1 CFS quotas, and system cores/RAM).
• Safe Defaults: Caps local FastEmbed (ONNX) CPU concurrency to min(2, detected_cpus) (default: 2 cores) and batch size to 32 chunks (down from FastEmbed's unconstrained 256) to eliminate container CPU spikes and host starvation during ingestion.
• Dynamic In-App Hot-Reloading: Added GET /admin/api/settings/embedding and POST /admin/api/settings/embedding allowing live modification of thread caps, batch sizes, and model configurations without restarts.
• Persistent Settings: Seeding and storage of embedding preferences in the SQLite system_metadata table.

🎨 Web Admin Dashboard Controls

• Embedding Engine & Resource Limits Panel: Dedicated card in Settings displaying active execution provider, live thread allocations vs. detected host hardware, batch size, and RAM capacity.
• Interactive Configuration Form: User-configurable CPU thread limit slider, batch size dropdown (16, 32, 64, 128, 256), dense/sparse model selectors, and LiteLLM API endpoint settings.

🧪 Test Coverage & Modularity

• Added comprehensive unit and component test suites (282 backend Pytest tests, 96 frontend Vitest tests).
• All source files strictly maintained under the sub-500 LOC modularity floor.
• Synchronized automated requirement specifications (REQUIREMENTS.md).

v2.9.0 - Multi-Theme Engine & Compact UI Overhaul

Highlights in v2.9.0:

• Multi-Theme Engine: Introduced 4 polished themes with zero-latency switching and localStorage persistence:
  • Deep Ocean (Dark Default): Petrol spruce palette with vibrant cyan and mint accents.
  • Midnight Blue (Dark Space): Classic obsidian space navy with royal blue and teal highlights.
  • Lavender Haze (Light Purple): Soft lilac-violet canvas with vibrant purple, fuchsia accents, and deep eggplant text.
  • Amber Warmth (Light Sandstone): Warm sandstone canvas with terracotta orange, amber accents, and dark espresso text.
• Appearance & Theme Settings Panel: Interactive theme switcher cards with live color palette swatches, dark/light mode badges, active checkmark indicators, and toast feedback.
• Card & Layout Compaction: Compacted glass-card padding, overview metric cards, icons, and theme selector grid footprint for a cleaner, higher-density dashboard experience.
• Consolidated Typography Architecture: Centralized font family, weight, and size definitions into CSS custom variables.
• Test Suite & Verification: Added comprehensive unit tests in ThemeSettings.test.tsx covering initial render, click-to-switch behavior, toast messages, legacy alias resolution, and invalid fallback handling (100% test pass rate across 277 backend Pytest tests and 89 frontend Vitest tests).
• Documentation & Screenshots Gallery: Updated README.md with high-resolution theme showcase comparisons and refreshed settings preview.

v2.8.0 - Modular Codebase Architecture

Highlights in v2.8.0:

• Modular Codebase Restructuring: Broke down legacy monolithic files into cohesive subpackages under app/services/database/, app/services/chunking/, app/services/indexing/, app/services/topology/, app/mcp/handlers/, and app/api/routers/.
• Target File Size Compliance: All individual source files in app/ and frontend/src/ are now maintained under 450 lines of code.
• Removed Backward Compatibility Shims: Direct modular imports across the codebase with full test suite alignment.
• Verification: 100% test pass rate across 277 backend pytest tests and 82 frontend Vitest component tests.

v2.7.0: Graph Intelligence, Topology Explorer, API Contracts, and Architecture Synthesis

What's Changed in ContextCortex v2.7.0

🌐 1. Interactive Visual Codebase & Dependency Topology Explorer (#8)

• Added an interactive force-directed graph canvas in the Web Admin Dashboard for exploring file hierarchies, module clusters, AST symbol relationships, and API routes.
• Real-time pan, zoom, drag physics, node/edge type filtering, and search-to-focus autocomplete.
• Slide-over Inspector Drawer with code previews, AST signatures, line ranges, neighbor connections, and Git provider permalinks.
• Backend REST endpoints: GET /admin/api/graph/topology and GET /admin/api/graph/node-details.

🌲 2. AST-Based Call Graph Traversal & Relationship Tracing (#7)

• Tree-sitter extraction across 10+ languages capturing CALLS, IMPORTS, INHERITS, and IMPLEMENTS relationships into indexed SQLite ast_relationships tables.
• FastMCP tool trace_path: Cycle-safe BFS traversal supporting inbound (callers), outbound (callees), and bidirectional (both) exploration with depth clamping and Markdown tree formatting.

🔗 3. API Route Discovery & Cross-Service Endpoint Linking (#6)

• Framework-aware HTTP route extraction for FastAPI, Express, Flask, ASP.NET, and Spring Boot into api_routes.
• Client call-site discovery (fetch, axios, requests, HttpClient) into api_client_calls.
• FastMCP tools find_routes and find_api_callers for instant API contract discovery and cross-service consumer tracing.

🏛️ 4. Architecture Overview Synthesis & Architecture Decision Records (#5)

• FastMCP tool get_architecture: Synthesizes primary language distributions, application entry points, key hotspot modules, route summaries, and active ADRs within a concise token footprint (<1,000 tokens).
• FastMCP tool manage_adr: Full Architecture Decision Record lifecycle management (list, get, create, update, supersede) with automatic markdown file ingestion (docs/adr/*.md).

🧪 5. Testing & Verification Baseline

• 383 Automated Tests: 277 Pytest Backend + 82 Vitest Frontend + 24 Playwright E2E tests passing.
• 100% clean CI/CD pipeline.