The Lego Analogy
Individual, composable primitives you can use independently — or snap together to build production-grade AI systems. Each piece works standalone. Together, they form the HALETHEIA stack.
The Pieces
Use one, use some, or use all. No forced coupling.
HCP Protocol
The git-native standard for AI context management. The foundation everything else builds on.
Protocol · AGPL-3.0 · Available nowLiiADA Framework
Modular agent orchestration in Python. 13 swappable modules, hexagonal architecture, HCP-native.
Framework · Python · Launching May 2026aronly
CLI that filters and compresses command output before it reaches your LLM. −60-90% tokens.
CLI · Rust · Available nowbabuino
Secure sandbox for AI agent skill execution. KVM-based isolation — not containers, real hardware boundaries.
Security · Rust · Available nowcutufato
Verifiable audit trail for AI agent operations. Every decision logged, tamper-evident, queryable.
Audit · Python · Available nowxokito
Privacy shield for MCP tools. Keeps your secrets hidden while agents work with configuration files.
Privacy · Python · Available nowLego ≠ Lock-in
Each piece is useful on its own. You don't need the full stack. Use HCP with Cursor. Use aronly in any terminal. Use babuino in your existing pipeline.
When you need more
The Castle
Enterprise infrastructure on top of the Legos: governance UI, reactive gateway, hallucination-free RAG, and EU AI Act compliance tooling. For teams that need SLAs.
See Enterprise →