ACR (Adaptive Cognitive Runtime) is a Python package that provides a local-first, model-independent runtime for AI orchestration and autonomous agent execution. It supports running agents with local models such as those served by Ollama, implements the Model Context Protocol (MCP), and enables building complex cognitive workflows without vendor lock-in. Ideal for developers building offline-capable or privacy-focused AI agent systems.
acr-runtime is an AI & ML project. It focuses on orchestrating and running autonomous AI agents locally without depending on specific LLM providers or cloud services. It is built as an open-source project for developers. The project is open source (MIT). acr-runtime is available on the command line and API, and it can be self-hosted.
It is developed by hbkdad, and it first shipped in 2026. The project is developed in the open on GitHub with 70 commits in the last 90 days. Among its 5 catalogued features are local-first execution, model independence, and agent orchestration. It exposes integrations via an MCP server and a public API.
Summary written by a language model from the project’s public pages.
What PulseGate has recorded for this listing
Closest matches by what these projects do