Documentation
Get started with IOWarp's context engineering platform. Guides, API references, deployment instructions.
Quick Start
Get IOWarp running in minutes with Docker Compose. Perfect for local development and testing.
# Clone the repository
git clone https://github.com/iowarp/iowarp.git
cd iowarp
# Start with Docker Compose
pip install iowarp-clio
clio deploy --mode docker
# Verify installation
clio statusFor native installation or Spack-based builds, see the Getting Started guide.
Documentation Sections
Getting Started
Docker Compose setup, Spack installation, native build. Quick start tutorial with benchmarks.
SDK Reference
Hermes Shared Memory (Interprocess), Chimaera Runtime Modules, Context Transfer Engine, OMNI File Format.
Python API
CLIO Python API: iowarp.clio module, ContextInterface, AssimilationCtx, data placement.
Deployment
HPC cluster deployment with Spack, YAML configuration reference, performance tuning, worker threads.
CLIO Kit (MCP)
15+ MCP servers detailed reference. Installation for Claude Code, Cursor, VS Code.
FAQ
Troubleshooting TCP server errors, common issues, configuration tips.
Data Model
Context = data + metadata + provenance + quality metrics. Hierarchical tiers: Critical (GPU HBM), Working (NVMe), Reference (PFS), Archive (tape/cloud).
Context Structure
- • Data: Actual content (files, datasets, streams)
- • Metadata: Schema, attributes, annotations
- • Provenance: Origin, transformations, lineage
- • Quality Metrics: Completeness, accuracy, freshness
Storage Tiers
- • Critical: GPU HBM (fastest, smallest)
- • Working: NVMe SSDs (high-speed cache)
- • Reference: Parallel File System (persistent)
- • Archive: Tape/Cloud (long-term storage)
IOWarp automatically manages data placement across tiers based on access patterns, cost, and performance requirements. Agents request context, and the system intelligently moves data to the optimal tier for each operation.