██████╗ ██╗███████╗████████╗██╗██╗ ██╗ ██╗████████╗
██╔══██╗██║██╔════╝╚══██╔══╝██║██║ ██║ ██║╚══██╔══╝
██║ ██║██║███████╗ ██║ ██║██║ ██║█████╗██║ ██║
██║ ██║██║╚════██║ ██║ ██║██║ ██║╚════╝██║ ██║
██████╔╝██║███████║ ██║ ██║███████╗ ███████╗ ██║ ██║
╚═════╝ ╚═╝╚══════╝ ╚═╝ ╚═╝╚══════╝ ╚══════╝ ╚═╝ ╚═╝
Point Claude at any GitHub repo. Get back structured knowledge you can actually reuse.
A knowledge base where Claude does the reading, I just pick the repos.
Give Claude a GitHub URL and it reads the codebase over the web, traces each feature end-to-end, and writes two documents per feature:
| Layer | Audience | Purpose |
|---|---|---|
| Study doc | You | Plain-language deep-dive, ready to upload to NotebookLM |
| Build spec | Claude (future sessions) | Self-contained technical guide for reimplementing the feature in any new project |
Everything feeds into an interactive knowledge graph you can browse and search.
Open graph/graph.html in a browser.
Click any node → see the summary, links to the study doc, build spec, and origin repo.
| Domain | Feature | From |
|---|---|---|
| pipeline-orchestration | Autopilot — Scheduled Agent Work | multica |
| runtime | Unified Runtimes + CLI Auto-Detection | multica |
| agent-architecture | Agents as Teammates | multica |
| agent-architecture | Autonomous Execution Lifecycle + WS Streaming | multica |
| research-automation | Agentic Research Planning | scira |
| structured-extraction | Grounded Retrieval + Inline Citations | scira |
| ai-integration | Multi-Provider Model Gateway | scira |
| agent-architecture | Tool & Search-Mode Registry | scira |
| streaming | Resumable Streaming Search | scira |
| pipeline-orchestration | Scheduled Monitoring Agents (Lookouts) | scira |
| realtime | Debounced File Watcher | hazelnut |
| pipeline-orchestration | File Rules Engine | hazelnut |
| pipeline-orchestration | File Actions Executor | hazelnut |
| data-structures | MBQL — Metabase Query AST | metabase |
| pipeline-orchestration | Query Processor Middleware Pipeline | metabase |
| plugin-architecture | Multimethod Driver Abstraction | metabase |
| rendering | Visualization Auto-Selection | metabase |
| pipeline-orchestration | Declarative (Low-Code) CDK | airbyte |
| data-portability | Airbyte Protocol | airbyte |
| pipeline-orchestration | Incremental Sync & State | airbyte |
| web-extraction | Declarative HTTP Stream Stack | airbyte |
| code-generation | Connector Builder Test-Read | airbyte |
| web-extraction | Site URL Map | firecrawl |
| web-extraction | Agentic Browser Actions | firecrawl |
| research-automation | Deep Research Loop | firecrawl |
| content-synthesis | Generate llms.txt | firecrawl |
| payments | Credit Billing & Concurrency | firecrawl |
| credential-management | Keyless & x402 Access | firecrawl |
| web-extraction | Scrape Engine + Fallback Pipeline | firecrawl |
| pipeline-orchestration | Queue-Backed Crawl | firecrawl |
| structured-extraction | LLM Extract (map-reduce) | firecrawl |
| web-extraction | Web Search (+ scrape) | firecrawl |
| ai-integration | Citation-Grounded Chat | openpaper |
| content-preprocessing | PDF Ingestion Pipeline | openpaper |
| canvas-interaction | PDF Highlights & Annotations | openpaper |
| web-extraction | Corpus & Academic Search | openpaper |
| tts | Audio Overview (TTS) | openpaper |
| credential-management | Multi-Path Authentication | openpaper |
| adaptive-parsing | Adaptive Element Relocation | Scrapling |
| document-conversion | Converter Pipeline | markitdown |
| document-conversion | PDF Conversion | markitdown |
| document-conversion | Office Doc Conversion | markitdown |
| document-conversion | ZIP Archive Traversal | markitdown |
| plugin-architecture | Plugin System | markitdown |
| file-detection | Magika File Detection | markitdown |
| media-processing | Image + LLM Captioning | markitdown |
| media-processing | Audio Transcription | markitdown |
| web-extraction | HTML Web Conversion | markitdown |
| web-extraction | YouTube Extraction | markitdown |
| ai-integration | Azure Doc Intelligence | markitdown |
| agent-architecture | Agent Output Contract | last30days-skill |
| research-automation | Multi-Source Research Engine | last30days-skill |
| research-automation | Entity Resolution | last30days-skill |
| research-automation | Engagement Signal Ranking | last30days-skill |
| content-synthesis | Cross-Source Clustering | last30days-skill |
| credential-management | Multi-Tier Credentials | last30days-skill |
| ai-automation | AI Rules Engine | inbox-zero |
| ai-automation | AI Reply Drafting | inbox-zero |
| inbox-cleanup | Bulk Unsubscriber | inbox-zero |
| inbox-cleanup | Bulk Archiver | inbox-zero |
| email-platform | Email Provider Abstraction | inbox-zero |
| agent-architecture | Ordered Backend Routing | Agent-Reach |
| diagnostics | Channel Health Diagnostics | Agent-Reach |
| credential-management | Cookie Credential Extraction | Agent-Reach |
| agent-distribution | Agent-Driven Install | Agent-Reach |
| tts | ONNX TTS Pipeline | supertonic |
| tts | Flow-Matching Sampler | supertonic |
| tts | Expression Tags | supertonic |
| ai-workflow | Agentic Loop | open-design |
| ai-workflow | Agent CLI Integration | open-design |
| codegen | Design Artifact Generation | open-design |
| design-systems | Design Systems Library | open-design |
| infrastructure | BYOK Proxy | open-design |
| infrastructure | Local-First Architecture | open-design |
| plugin-architecture | Skills System | open-design |
| plugin-architecture | Plugin Ecosystem | open-design |
| reactivity | Signals Reactivity Engine | tldraw |
| state-management | Reactive Record Store | tldraw |
| schema-migrations | Schema & Migrations | tldraw |
| realtime | Multiplayer Sync | tldraw |
| design-systems | Token Pipeline Orchestration | style-dictionary |
| design-systems | Reference Resolution Engine | style-dictionary |
| design-systems | Transforms & Transform Groups | style-dictionary |
| plugin-architecture | Register / Extensibility API | style-dictionary |
| canvas-interaction | Pan & Zoom Canvas | xyflow |
| canvas-interaction | Node Dragging | xyflow |
| canvas-interaction | Minimap Navigation | xyflow |
| graph-editing | Connection Handles | xyflow |
| graph-editing | Node Resizer | xyflow |
| graph-rendering | Edge Path Algorithms | xyflow |
| state-management | Reactive Store Architecture | xyflow |
| rendering | Hand-Drawn Rendering | excalidraw |
| realtime-collab | E2E-Encrypted Collaboration | excalidraw |
| realtime-collab | Scene Reconciliation | excalidraw |
| data-structures | Fractional Indexing (z-order) | excalidraw |
| code-generation | Interview-Driven App Scaffolding | whatsapp-agentkit |
| messaging | WhatsApp Provider Adapter Layer | whatsapp-agentkit |
| agent-architecture | Per-Contact Conversation Memory | whatsapp-agentkit |
| lead-scoring | Rule-Based Lead Scoring | auto-crm |
| ai-integration | AI Lead Classification (Claude) | auto-crm |
| lead-ingestion | Webhook Lead Ingestion | auto-crm |
| agent-architecture | MCP CRM Server | auto-crm |
| canvas-interaction | Kanban Pipeline (drag-and-drop) | auto-crm |
| analytics | CRM Dashboard KPIs | auto-crm |
| data-portability | CSV Import / Export | auto-crm |
| activity-tracking | Activity Tracking & Follow-up Buckets | auto-crm |
| agent-distribution | Self-Customizing CRM (Claude commands) | auto-crm |
| content-synthesis | AI Carousel Generation | carousel-generator |
| rendering | DOM-to-PDF Carousel Export | carousel-generator |
| design-systems | OKLCH Theme Palettes | carousel-generator |
| data-portability | Zod Form Persistence & JSON Portability | carousel-generator |
| infrastructure | BYOK + Rate-Limited AI Action | carousel-generator |
| pipeline-orchestration | Graph Execution Engine | scrapegraph-ai |
| web-extraction | SmartScraper Pipeline | scrapegraph-ai |
| web-extraction | Multi-Source Fetch Node | scrapegraph-ai |
| structured-extraction | Map-Reduce Answer Generation | scrapegraph-ai |
| ai-integration | Provider-Agnostic Model Layer | scrapegraph-ai |
| research-automation | Search-Driven Scraping | scrapegraph-ai |
| state-management | Change-Based Mutation Model | penpot |
| design-systems | Native Design Tokens | penpot |
| rendering | WASM/Skia Render Engine | penpot |
| macos-ui | Notch-Shaped Always-On-Top Window | boring.notch |
| media-control | Multi-Provider Media Control | boring.notch |
| macos-ui | System HUD Replacement | boring.notch |
| ai-workflow | Chat-Completion Middleware | open-webui |
| permissions | Unified Permission Abstraction | PermissionsKit |
| infrastructure | Modular Per-Permission Packaging | PermissionsKit |
| infrastructure | Async-to-Sync Status Bridging | PermissionsKit |
| media-processing | Duration-Aware Frame Budgeting | claude-video |
| transcription | Captions-First Transcription Cascade | claude-video |
| agent-architecture | Agentic Video-Understanding Pipeline (/watch) | claude-video |
| storage-engine | In-Memory MVCC Storage Engine | memgraph |
| durability | WAL + Snapshot Durability | memgraph |
| vector-search | USearch Vector Index | memgraph |
| power-management | Power-Assertion Wake Lock | open-caffeine |
| app-distribution | Sparkle EdDSA Auto-Update | open-caffeine |
| testing-discipline | Coverage-Gated Testable Core | open-caffeine |
| search-relevance | Ranking-Rules Bucket Sort | meilisearch |
| fuzzy-search | Length-Gated Typo Tolerance | meilisearch |
| task-scheduling | Auto-Batching Write Scheduler | meilisearch |
| agent-architecture | MCP Server for a Native Editor | palmier-pro |
| agent-architecture | Intent-Based Editor Tools with Shared Undo | palmier-pro |
| video-editing | Transcript-Driven Clip Editing | openchatcut |
| transcription | Local-First, Multi-Provider ASR Pipeline | openchatcut |
| agent-architecture | Propose -> Review -> Apply Agent Edits | openchatcut |
| agent-architecture | Project-Bound MCP Draft Sessions | openchatcut |
distill-it/
├── graph/
│ ├── graph.json ← single source of truth (all nodes + edges)
│ └── graph.html ← open in browser for the interactive map
├── features/ ← organized by WHAT it does, not which repo
│ └── <domain>/
│ ├── _domain.md
│ ├── study/ ← human-readable, NotebookLM-ready
│ └── build/ ← agent-ready transplant spec
└── repos/
└── <repo>.md ← index of everything extracted from one repo
Features are filed by what they do — so all "file detection" approaches sit together regardless of which repo they came from.
Run /distill https://github.com/owner/repo in Claude Code.
Claude reads the repo over the web (no cloning), shows you the feature list, writes the study + build docs for whichever you pick, updates the graph, commits, and pushes.
Built with Claude Code using the repository-intelligence skill.