HCFS - Hierarchical Context File System
Context-Aware Hierarchical Context File System (HCFS): Unifying file system paths with context blobs for agentic AI cognition
Overview
HCFS is a virtual filesystem layer that maps hierarchical paths to context blobs, enabling agentic AI systems to navigate and share context in a structured, hierarchical manner. It combines the intuitive nature of file system navigation with semantic context storage and retrieval.
Key Features
- Virtual Filesystem Layer: Standard POSIX-style directory navigation backed by context blobs
- Context Database Backend: Versioned context storage with hierarchical inheritance
- Semantic Indexing: Embeddings and BM25 hybrid ranking for context relevance
- Agent APIs: Syscall-style APIs for context navigation, retrieval, and publishing
- Decentralized Context Sharing: Agents can publish/subscribe to context updates by path
Quick Start
This project is currently in the planning and research phase. See PROJECT_PLAN.md for detailed architecture and implementation timeline.
Architecture
┌─────────────────┐ ┌─────────────────┐ ┌─────────────────┐
│ Agent APIs │ │ Virtual FS │ │ Context DB │
│ │ │ Layer (FUSE) │ │ Backend │
│ • context_cd() │◄──►│ │◄──►│ │
│ • context_get() │ │ /project/ │ │ • Blob storage │
│ • context_push()│ │ /project/src/ │ │ • Versioning │
│ • context_list()│ │ /project/docs/ │ │ • Embeddings │
└─────────────────┘ └─────────────────┘ └─────────────────┘
Development Phases
- Phase 0: Research & Design (2 weeks)
- Phase 1: Prototype FS layer (4 weeks)
- Phase 2: Backend DB & storage (4 weeks)
- Phase 3: Embedding & retrieval integration (3 weeks)
- Phase 4: API/Syscall layer scripting (3 weeks)
- Phase 5: Agent integration & simulation (3 weeks)
- Phase 6: Evaluation & refinement (2 weeks)
- Phase 7: Write-up & publication (2 weeks)
Contributing
This project is in early development. See PROJECT_PLAN.md for detailed specifications and implementation roadmap.
License
MIT License - see LICENSE for details.
Research Context
HCFS builds upon research in semantic file systems, LLM-driven semantic filesystems (LSFS), path-structure embeddings, and context modeling frameworks. See the literature review section in PROJECT_PLAN.md for full references.