Show HN: Agx – A Kanban board that runs your AI coding agents (github.com)

AGX has launched a local-first agent orchestrator designed to enhance the functionality of AI coding CLIs (such as Claude Code, Gemini, and Ollama) by transforming them into autonomous agents capable of running uninterrupted across sessions. Unlike traditional systems that rely on conversational memory, AGX stores the authoritative state in a local PostgreSQL database, allowing agents to resume tasks seamlessly after crashes or restarts. This innovative approach enables a "Wake → Work → Sleep" cycle, where agents load context directly, execute tasks, and checkpoint their state, ensuring reliability and reducing overhead during long-running tasks. This development is significant for the AI/ML community as it addresses the common challenges of memory management and state persistence in automated coding agents. AGX's architecture separates the execution state from history, allowing the system to maintain a constant-cost operation regardless of task duration. Moreover, the solution runs entirely on users' machines, providing full observability and control over agent activities without reliance on hosted services or prompt-replay methodologies. As a result, users can define custom workflows and manage tasks with considerable ease using a bundled Kanban-style dashboard, fostering a more efficient development environment tailored to the needs of modern software engineering.