Show HN: Aigent-OS, a Claude Code harness that clears and resumes (github.com)

Aigent-OS has been announced as an innovative open-source harness for Claude Code, designed to enhance long-running AI sessions by implementing a Managed Auto-Refresh feature. Instead of the conventional methods of context management—like auto-compacting or manual clears—Aigent-OS automatically saves a verified capsule of the current context before it reaches its threshold, executes a single /clear command, and resumes work in a fresh session. This ensures continuity without the need for repeated briefings, which can disrupt workflow and efficiency. This development is significant for the AI/ML community as it addresses common limitations tied to Claude Code's fixed context window, allowing for a more efficient interaction pattern with AI. Key technical components include configurable thresholds for capsule creation and an operator layer that manages skill routing, memory retention, and specialist sub-agents for various tasks. With sessions running on user machines and memory stored as plain markdown in git, this solution promises a more transparent and manageable approach to AI interactions, making it easier to track decisions and uncompleted tasks directly within the user’s workflow. Although currently validated through a single-seat setup, upcoming improvements and fixes aim to expand its reliability and functionality.