A lightweight, production-grade execution kernel for managing the LLM interaction lifecycle.
Governed execution kernel for AI agents. Default-deny policy, bounded capabilities, typed execution, transactional Git. Evidence over claims.
An execution kernel that treats LLMs as untrusted compute and enforces policy via deterministic runtime interception.
Safe, auditable execution kernel for AI agent tools — DDD in Go, zero dependencies, with approval gates, execution budgets, evidence trails, and typed contracts.
SomaOS Brain Next - Zero-dependency motor execution kernel for SO-100 6-DoF arm. Dual-layer action filter, VLA hot-plug, high-value failure replay, 1GB/4-core ARM ceiling.
To associate your repository with the execution-kernel topic, visit your repo's landing page and select "manage topics."